agora inbox for pgsql-hackers@postgresql.org
help / color / mirror / Atom feed[PATCH v1 4/5] Allow to resize shared memory without restart
425+ messages / 2 participants
[nested] [flat]
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v1 4/5] Allow to resize shared memory without restart
@ 2024-10-16 18:24 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2024-10-16 18:24 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Size for every shared memory slot is recalculated based on the new
NBuffers, and extended using mremap. After allocating new space, new
shared structures (buffer blocks, descriptors, etc) are allocated as
needed. Here is how it looks like after raising shared_buffers from 128
MB to 512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v2 5/6] Allow to resize shared memory without restart
@ 2025-02-20 20:12 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-02-20 20:12 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a global
Barrier to coordinate backends that simultaneously change
shared_buffers, and pieces in shared memory to coordinate backends that
are too late to the party for some reason.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier. Afterwards it does shared memory
resize itself.
* All the backends, that participate in ProcSignal mechanism,
recalculate shared memory size based on the new NBuffers and extend it
using mremap.
* When finished, a backend waits on a global ShmemControl barrier,
untill all backends will be finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all backends are using new value, one backend will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f5a2bd04000-7f5a32e52000 /dev/zero (deleted)
7f5a39252000-7f5a4030e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d7ba000 /dev/zero (deleted)
7f5a53bba000-7f5a5ad26000 /dev/zero (deleted)
7f5a9ad26000-7f5aa9d94000 /dev/zero (deleted)
^ buffers mapping, ~240 MB
7f5d29d94000-7f5d30e00000 /dev/zero (deleted)
-- 512 MB
7f5a2bd04000-7f5a33274000 /dev/zero (deleted)
7f5a39252000-7f5a4057e000 /dev/zero (deleted)
7f5a4670e000-7f5a4d9fa000 /dev/zero (deleted)
7f5a53bba000-7f5a5b1a6000 /dev/zero (deleted)
7f5a9ad26000-7f5ac1f14000 /dev/zero (deleted)
^ buffers mapping, ~625 MB
7f5d29d94000-7f5d30f80000 /dev/zero (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v4 6/8] Allow to resize shared memory without restart
@ 2025-04-06 14:47 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-04-06 14:47 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers and extend it using mremap. One elected process signals the
postmaster to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f90cde00000-7f90d4fa6000 /dev/zero (deleted)
7f90d4fa6000-7f914de00000
7f914de00000-7f915cfa8000 /dev/zero (deleted)
^ buffers mapping, ~241 MB
7f915cfa8000-7f944de00000
7f944de00000-7f94550a8000 /dev/zero (deleted)
7f94550a8000-7f94cde00000
7f94cde00000-7f94d4fe8000 /dev/zero (deleted)
7f94d4fe8000-7f954de00000
7f954de00000-7f9554ff6000 /dev/zero (deleted)
7f9554ff6000-7f958de00000
7f958de00000-7f959508a000 /dev/zero (deleted)
7f959508a000-7f95cde00000
-- 512 MB
7f90cde00000-7f90d5126000 /dev/zero (deleted)
7f90d5126000-7f914de00000
7f914de00000-7f9175128000 /dev/zero (deleted)
^ buffers mapping, ~627 MB
7f9175128000-7f944de00000
7f944de00000-7f9455528000 /dev/zero (deleted)
7f9455528000-7f94cde00000
7f94cde00000-7f94d5228000 /dev/zero (deleted)
7f94d5228000-7f954de00000
7f954de00000-7f9555266000 /dev/zero (deleted)
7f9555266000-7f958de00000
7f958de00000-7f95954aa000 /dev/zero (deleted)
7f95954aa000-7f95cde00000
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH v5 07/10] Allow to resize shared memory without restart
@ 2025-06-17 12:16 Dmitrii Dolgov <9erthalion6@gmail.com>
0 siblings, 0 replies; 425+ messages in thread
From: Dmitrii Dolgov @ 2025-06-17 12:16 UTC (permalink / raw)
Add assing hook for shared_buffers to resize shared memory using space,
introduced in the previous commits without requiring PostgreSQL restart.
Essentially the implementation is based on two mechanisms: a
ProcSignalBarrier is used to make sure all processes are starting the
resize procedure simultaneously, and a global Barrier is used to
coordinate after that and make sure all finished processes are waiting
for others that are in progress.
The resize process looks like this:
* The GUC assign hook sets a flag to let the Postmaster know that resize
was requested.
* Postmaster verifies the flag in the event loop, and starts the resize
by emitting a ProcSignal barrier.
* All processes, that participate in ProcSignal mechanism, begin to
process ProcSignal barrier. First a process waits until all processes
have confirmed they received the message and can start simultaneously.
* Every process recalculates shared memory size based on the new
NBuffers, adjusts its size using ftruncate and adjust reservation
permissions with mprotect. One elected process signals the postmaster
to do the same.
* When finished, every process waits on a global ShmemControl barrier,
untill all others are finished as well. This way we ensure three
stages with clear boundaries: before the resize, when all processes
use old NBuffers; during the resize, when processes have mix of old
and new NBuffers, and wait until it's done; after the resize, when all
processes use new NBuffers.
* After all processes are using new value, one of them will initialize
new shared structures (buffer blocks, descriptors, etc) as needed and
broadcast new value of NBuffers via ShmemControl in shared memory.
Other backends are waiting for this operation to finish as well. Then
the barrier is lifted and everything goes as usual.
Since resizing takes time, we need to take into account that during that time:
- New backends can be spawned. They will check status of the barrier
early during the bootstrap, and wait until everything is over to work
with the new NBuffers value.
- Old backends can exit before attempting to resize. Synchronization
used between backends relies on ProcSignalBarrier and waits for all
participants received the message at the beginning to gather all
existing backends.
- Some backends might be blocked and not responsing either before or
after receiving the message. In the first case such backend still
have ProcSignalSlot and should be waited for, in the second case
shared barrier will make sure we still waiting for those backends. In
any case there is an unbounded wait.
- Backends might join barrier in disjoint groups with some time in
between. That means that relying only on the shared dynamic barrier is
not enough -- it will only synchronize resize procedure withing those
groups. That's why we wait first for all participants of ProcSignal
mechanism who received the message.
Here is how it looks like after raising shared_buffers from 128 MB to
512 MB and calling pg_reload_conf():
-- 128 MB
7f87909fc000-7f8798248000 rw-s /memfd:strategy (deleted)
7f8798248000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a4e84000 rw-s /memfd:checkpoint (deleted)
7f87a4e84000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1b42000 rw-s /memfd:iocv (deleted)
7f87b1b42000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cb59c000 rw-s /memfd:descriptors (deleted)
7f87cb59c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f87ece38000 rw-s /memfd:buffers (deleted)
^ buffers content, ~247 MB
7f87ece38000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~2210 MB
7f8877066000-7f887e7d0000 rw-s /memfd:main (deleted)
7f887e7d0000-7f8890a00000 ---s /memfd:main (deleted)
-- 512 MB
7f87909fc000-7f879866a000 rw-s /memfd:strategy (deleted)
7f879866a000-7f879d6ca000 ---s /memfd:strategy (deleted)
7f879d6ca000-7f87a50f4000 rw-s /memfd:checkpoint (deleted)
7f87a50f4000-7f87aa398000 ---s /memfd:checkpoint (deleted)
7f87aa398000-7f87b1d82000 rw-s /memfd:iocv (deleted)
7f87b1d82000-7f87c3d32000 ---s /memfd:iocv (deleted)
7f87c3d32000-7f87cba1c000 rw-s /memfd:descriptors (deleted)
7f87cba1c000-7f87dd6cc000 ---s /memfd:descriptors (deleted)
7f87dd6cc000-7f8804fb8000 rw-s /memfd:buffers (deleted)
^ buffers content, ~632 MB
7f8804fb8000-7f8877066000 ---s /memfd:buffers (deleted)
^ reserved space, ~1824 MB
7f8877066000-7f887e950000 rw-s /memfd:main (deleted)
7f887e950000-7f8890a00000 ---s /memfd:main (deleted)
The implementation supports only increasing of shared_buffers. For
decreasing the value a similar procedure is needed. But the buffer
blocks with data have to be drained first, so that the actual data set
fits into the new smaller space.
^ permalink raw reply [nested|flat] 425+ messages in thread
* [PATCH 5/5] Serialize decoded tuples without flattening.
@ 2026-02-27 18:01 Antonin Houska <ah@cybertec.at>
0 siblings, 0 replies; 425+ messages in thread
From: Antonin Houska @ 2026-02-27 18:01 UTC (permalink / raw)
So far we used to flatten the tuples produced by logical decoding in order to
write them to a temporary file. That means that all the TOASTed attributes
ended up in a single chunk of memory. The problem is that even if individual
attributes do fit into the maximum chunk size (currently 1 GB), the whole
tuple is not guaranteed to fit.
The flattening might be a problem in the future anyway, since there's an
effort to restrict the impact of REPACK (CONCURRENTLY) on the VACUUM xmin
horizon. However, as the flattening assumes scans of the TOAST relaation, the
less we restrict VACUUM, the more likely it is that the TOAST attributes
needed for the flattening won't be available at decoding time.
(This patch also removes unnecessary field 'tupdesc' from RepackDecodingState
and allocates it in the appropriate memory context.)
---
src/backend/commands/cluster.c | 260 +++++++++++++++---
.../pgoutput_repack/pgoutput_repack.c | 114 ++++++--
src/include/commands/cluster.h | 5 +-
.../expected/repack_toast.out | 1 +
.../injection_points/specs/repack_toast.spec | 7 +
5 files changed, 318 insertions(+), 69 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index db710259eeb..e4de410e733 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -35,6 +35,7 @@
#include "postgres.h"
#include "access/amapi.h"
+#include "access/detoast.h"
#include "access/heapam.h"
#include "access/multixact.h"
#include "access/relscan.h"
@@ -285,6 +286,10 @@ static void apply_concurrent_update(Relation rel, HeapTuple tup,
IndexInsertState *iistate,
TupleTableSlot *index_slot);
static void apply_concurrent_delete(Relation rel, HeapTuple tup_target);
+static HeapTuple restore_tuple(BufFile *file, Relation relation,
+ MemoryContext cxt);
+static HeapTuple adjust_toast_pointers(Relation relation, HeapTuple tup_dst,
+ HeapTuple tup_src, MemoryContext cxt);
static HeapTuple find_target_tuple(Relation rel, ChangeDest *dest,
HeapTuple tup_key,
TupleTableSlot *ident_slot);
@@ -2574,6 +2579,7 @@ setup_logical_decoding(Oid relid)
LogicalDecodingContext *ctx;
NameData slotname;
RepackDecodingState *dstate;
+ MemoryContext oldcxt;
/*
* REPACK CONCURRENTLY is not allowed in a transaction block, so this
@@ -2624,14 +2630,6 @@ setup_logical_decoding(Oid relid)
*/
ctx->reader->routine.page_read = read_local_xlog_page_no_wait;
- /*
- * read_local_xlog_page_no_wait() needs to be able to indicate the end of
- * WAL.
- */
- ctx->reader->private_data = MemoryContextAllocZero(ctx->context,
- sizeof(ReadLocalXLogPageNoWaitPrivate));
-
-
/* Some WAL records should have been read. */
Assert(ctx->reader->EndRecPtr != InvalidXLogRecPtr);
@@ -2642,18 +2640,23 @@ setup_logical_decoding(Oid relid)
XLByteToSeg(ctx->reader->EndRecPtr, repack_current_segment,
wal_segment_size);
- dstate = palloc0_object(RepackDecodingState);
- dstate->relid = relid;
+ /* Our private state belongs to the decoding context. */
+ oldcxt = MemoryContextSwitchTo(ctx->context);
/*
- * Tuple descriptor may be needed to flatten a tuple before we write it to
- * a file. A copy is needed because the decoding worker invalidates system
- * caches before it starts to do the actual work.
+ * read_local_xlog_page_no_wait() needs to be able to indicate the end of
+ * WAL.
*/
- rel = table_open(relid, AccessShareLock);
- dstate->tupdesc = CreateTupleDescCopy(RelationGetDescr(rel));
+ ctx->reader->private_data = palloc0_object(ReadLocalXLogPageNoWaitPrivate);
+ dstate = palloc0_object(RepackDecodingState);
+ MemoryContextSwitchTo(oldcxt);
+
+#ifdef USE_ASSERT_CHECKING
+ dstate->relid = relid;
+#endif
/* Avoid logical decoding of other relations. */
+ rel = table_open(relid, AccessShareLock);
repacked_rel_locator = rel->rd_locator;
toastrelid = rel->rd_rel->reltoastrelid;
if (OidIsValid(toastrelid))
@@ -2822,10 +2825,10 @@ static void
apply_concurrent_changes(BufFile *file, ChangeDest *dest)
{
char kind;
- uint32 t_len;
Relation rel = dest->rel;
TupleTableSlot *index_slot,
*ident_slot;
+ MemoryContext apply_cxt;
HeapTuple tup_old = NULL;
/* TupleTableSlot is needed to pass the tuple to ExecInsertIndexTuples(). */
@@ -2835,6 +2838,14 @@ apply_concurrent_changes(BufFile *file, ChangeDest *dest)
/* A slot to fetch tuples from identity index. */
ident_slot = table_slot_create(rel, NULL);
+ /*
+ * Use a separate memory context for the tuples and any memory needed to
+ * process them.
+ */
+ apply_cxt = AllocSetContextCreate(TopTransactionContext,
+ "REPACK - apply",
+ ALLOCSET_DEFAULT_SIZES);
+
while (true)
{
size_t nread;
@@ -2849,14 +2860,7 @@ apply_concurrent_changes(BufFile *file, ChangeDest *dest)
break;
/* Read the tuple. */
- BufFileReadExact(file, &t_len, sizeof(t_len));
- tup = (HeapTuple) palloc(HEAPTUPLESIZE + t_len);
- tup->t_data = (HeapTupleHeader) ((char *) tup + HEAPTUPLESIZE);
- BufFileReadExact(file, tup->t_data, t_len);
- tup->t_len = t_len;
- ItemPointerSetInvalid(&tup->t_self);
- tup->t_tableOid = RelationGetRelid(dest->rel);
-
+ tup = restore_tuple(file, rel, apply_cxt);
if (kind == CHANGE_UPDATE_OLD)
{
Assert(tup_old == NULL);
@@ -2868,7 +2872,7 @@ apply_concurrent_changes(BufFile *file, ChangeDest *dest)
apply_concurrent_insert(rel, tup, dest->iistate, index_slot);
- pfree(tup);
+ MemoryContextReset(apply_cxt);
}
else if (kind == CHANGE_UPDATE_NEW || kind == CHANGE_DELETE)
{
@@ -2892,18 +2896,28 @@ apply_concurrent_changes(BufFile *file, ChangeDest *dest)
elog(ERROR, "failed to find target tuple");
if (kind == CHANGE_UPDATE_NEW)
+ {
+ /*
+ * If 'tup' contains TOAST pointers, they point to the old
+ * relation's toast. Copy the corresponding TOAST pointers for
+ * the new relation from the existing tuple. (The fact that we
+ * received a TOAST pointer here implies that the attribute
+ * hasn't changed.)
+ */
+ if (HeapTupleHasExternal(tup))
+ tup = adjust_toast_pointers(rel, tup, tup_exist,
+ apply_cxt);
+
apply_concurrent_update(rel, tup, tup_exist, dest->iistate,
index_slot);
+ }
else
apply_concurrent_delete(rel, tup_exist);
if (tup_old != NULL)
- {
- pfree(tup_old);
tup_old = NULL;
- }
- pfree(tup);
+ MemoryContextReset(apply_cxt);
}
else
elog(ERROR, "unrecognized kind of change: %d", kind);
@@ -2920,6 +2934,7 @@ apply_concurrent_changes(BufFile *file, ChangeDest *dest)
}
/* Cleanup. */
+ MemoryContextDelete(apply_cxt);
ExecDropSingleTupleTableSlot(index_slot);
ExecDropSingleTupleTableSlot(ident_slot);
}
@@ -3030,6 +3045,186 @@ apply_concurrent_delete(Relation rel, HeapTuple tup_target)
pgstat_progress_incr_param(PROGRESS_REPACK_HEAP_TUPLES_DELETED, 1);
}
+/*
+ * Read tuple from file and store it in given memory context.
+ *
+ * External attributes are stored in separate memory chunks, in order to avoid
+ * exceeding MaxAllocSize - that could happen if the individual attributes are
+ * smaller than MaxAllocSize but the whole tuple is bigger.
+ */
+static HeapTuple
+restore_tuple(BufFile *file, Relation relation, MemoryContext cxt)
+{
+ uint32 t_len;
+ HeapTuple tup;
+ MemoryContext oldcxt;
+ uint32 natt_ext;
+ List *attrs_ext = NIL;
+
+ oldcxt = MemoryContextSwitchTo(cxt);
+
+ /* First, read the external attributes if there are some. */
+ BufFileReadExact(file, &natt_ext, sizeof(natt_ext));
+ for (int i = 0; i < natt_ext; i++)
+ {
+ varlena varhdr;
+ char *ext_val;
+ Size ext_val_size;
+
+ /*
+ * Read the header into properly aligned memory before accessing it.
+ */
+ BufFileReadExact(file, &varhdr, VARHDRSZ);
+ ext_val_size = VARSIZE_ANY(&varhdr);
+ ext_val = palloc0(ext_val_size);
+ memcpy(ext_val, &varhdr, VARHDRSZ);
+ BufFileReadExact(file, ext_val + VARHDRSZ, ext_val_size - VARHDRSZ);
+
+ attrs_ext = lappend(attrs_ext, ext_val);
+ }
+
+ /* Read the tuple. */
+ BufFileReadExact(file, &t_len, sizeof(t_len));
+ tup = (HeapTuple) palloc(HEAPTUPLESIZE + t_len);
+ tup->t_data = (HeapTupleHeader) ((char *) tup + HEAPTUPLESIZE);
+ BufFileReadExact(file, tup->t_data, t_len);
+ tup->t_len = t_len;
+ ItemPointerSetInvalid(&tup->t_self);
+ tup->t_tableOid = RelationGetRelid(relation);
+
+ /*
+ * If there are "external indirect" attributes, make the tuple reference
+ * them.
+ */
+ if (natt_ext > 0)
+ {
+ TupleDesc desc;
+ Datum *attrs;
+ bool *isnull;
+ ListCell *lc;
+
+ desc = RelationGetDescr(relation);
+ attrs = palloc0_array(Datum, desc->natts);
+ isnull = palloc0_array(bool, desc->natts);
+
+ heap_deform_tuple(tup, desc, attrs, isnull);
+
+ lc = list_head(attrs_ext);
+ for (int i = 0; i < desc->natts; i++)
+ {
+ CompactAttribute *attr = TupleDescCompactAttr(desc, i);
+ varlena *varlena_pointer,
+ *new_datum;
+ varatt_indirect redirect_pointer;
+
+ /* Find the external attributes, like in store_change(). */
+ if (attr->attisdropped)
+ continue;
+ if (attr->attlen != -1)
+ continue;
+ if (isnull[i])
+ continue;
+ varlena_pointer = (varlena *) DatumGetPointer(attrs[i]);
+ if (!VARATT_IS_EXTERNAL_INDIRECT(varlena_pointer))
+ continue;
+
+ /* Update the attribute. */
+ memset(&redirect_pointer, 0, sizeof(redirect_pointer));
+ redirect_pointer.pointer = lfirst(lc);
+
+ new_datum = (varlena *) palloc0(INDIRECT_POINTER_SIZE);
+ SET_VARTAG_EXTERNAL(new_datum, VARTAG_INDIRECT);
+ memcpy(VARDATA_EXTERNAL(new_datum), &redirect_pointer,
+ sizeof(redirect_pointer));
+
+ attrs[i] = PointerGetDatum(new_datum);
+
+ lc = lnext(attrs_ext, lc);
+ }
+ Assert(lc == NULL);
+
+ tup = heap_form_tuple(desc, attrs, isnull);
+ }
+
+ MemoryContextSwitchTo(oldcxt);
+
+ return tup;
+}
+
+/*
+ * Copy TOAST pointers from 'tup_src' to 'tup_dst' and return the modified
+ * 'tup_dst'. Use 'cxt' for all memory allocations.
+ */
+static HeapTuple
+adjust_toast_pointers(Relation relation, HeapTuple tup_dst,
+ HeapTuple tup_src, MemoryContext cxt)
+{
+ TupleDesc desc;
+ Datum *attrs_dst;
+ bool *isnull_dst;
+ Datum *attrs_src = NULL;
+ bool *isnull_src = NULL;
+ MemoryContext oldcxt;
+ HeapTuple result;
+
+ oldcxt = MemoryContextSwitchTo(cxt);
+
+ desc = RelationGetDescr(relation);
+ attrs_dst = palloc0_array(Datum, desc->natts);
+ isnull_dst = palloc0_array(bool, desc->natts);
+
+ heap_deform_tuple(tup_dst, desc, attrs_dst, isnull_dst);
+
+ for (int i = 0; i < desc->natts; i++)
+ {
+ CompactAttribute *attr = TupleDescCompactAttr(desc, i);
+ varlena *varlena_dst,
+ *varlena_src;
+
+ if (attr->attisdropped)
+ continue;
+ if (attr->attlen != -1)
+ continue;
+ if (isnull_dst[i])
+ continue;
+
+ varlena_dst = (varlena *) DatumGetPointer(attrs_dst[i]);
+ if (!VARATT_IS_EXTERNAL_ONDISK(varlena_dst))
+ continue;
+
+ /*
+ * The first TOAST pointer? The caller can easily check if the tuple
+ * has external attributes(s) in general, but is not supposed to check
+ * specifically for TOAST.
+ */
+ if (attrs_src == NULL)
+ {
+ attrs_src = palloc0_array(Datum, desc->natts);
+ isnull_src = palloc0_array(bool, desc->natts);
+
+ heap_deform_tuple(tup_src, desc, attrs_src, isnull_src);
+ }
+
+ varlena_src = (varlena *) DatumGetPointer(attrs_src[i]);
+
+ /*
+ * Logical decoding only uses TOAST pointer for attributes that
+ * haven't changed.
+ */
+ Assert(!isnull_src[i]);
+ Assert(VARATT_IS_EXTERNAL_ONDISK(varlena_src));
+
+ /* Copy the pointer. */
+ memcpy(VARDATA_EXTERNAL(varlena_dst), VARDATA_EXTERNAL(varlena_src),
+ sizeof(varatt_external));
+ }
+ result = heap_form_tuple(desc, attrs_dst, isnull_dst);
+
+ MemoryContextSwitchTo(oldcxt);
+
+ return result;
+}
+
/*
* Find the tuple to be updated or deleted.
*
@@ -3263,15 +3458,8 @@ free_index_insert_state(IndexInsertState *iistate)
static void
cleanup_logical_decoding(LogicalDecodingContext *ctx)
{
- RepackDecodingState *dstate;
-
- dstate = (RepackDecodingState *) ctx->output_writer_private;
-
- FreeTupleDesc(dstate->tupdesc);
FreeDecodingContext(ctx);
-
ReplicationSlotDropAcquired();
- pfree(dstate);
}
/*
diff --git a/src/backend/replication/pgoutput_repack/pgoutput_repack.c b/src/backend/replication/pgoutput_repack/pgoutput_repack.c
index d0c464a98d5..8e22aeb163b 100644
--- a/src/backend/replication/pgoutput_repack/pgoutput_repack.c
+++ b/src/backend/replication/pgoutput_repack/pgoutput_repack.c
@@ -12,7 +12,7 @@
*/
#include "postgres.h"
-#include "access/heaptoast.h"
+#include "access/detoast.h"
#include "commands/cluster.h"
#include "replication/snapbuild.h"
#include "utils/memutils.h"
@@ -28,7 +28,7 @@ static void plugin_commit_txn(LogicalDecodingContext *ctx,
ReorderBufferTXN *txn, XLogRecPtr commit_lsn);
static void plugin_change(LogicalDecodingContext *ctx, ReorderBufferTXN *txn,
Relation rel, ReorderBufferChange *change);
-static void store_change(LogicalDecodingContext *ctx,
+static void store_change(LogicalDecodingContext *ctx, Relation relation,
ConcurrentChangeKind kind, HeapTuple tuple);
void
@@ -97,9 +97,8 @@ plugin_change(LogicalDecodingContext *ctx, ReorderBufferTXN *txn,
dstate = (RepackDecodingState *) ctx->output_writer_private;
- /* Only interested in one particular relation. */
- if (relation->rd_id != dstate->relid)
- return;
+ /* Changes of other relation should not have been decoded. */
+ Assert(RelationGetRelid(relation) == dstate->relid);
/* Decode entry depending on its type */
switch (change->action)
@@ -117,7 +116,7 @@ plugin_change(LogicalDecodingContext *ctx, ReorderBufferTXN *txn,
if (newtuple == NULL)
elog(ERROR, "incomplete insert info.");
- store_change(ctx, CHANGE_INSERT, newtuple);
+ store_change(ctx, relation, CHANGE_INSERT, newtuple);
}
break;
case REORDER_BUFFER_CHANGE_UPDATE:
@@ -132,9 +131,9 @@ plugin_change(LogicalDecodingContext *ctx, ReorderBufferTXN *txn,
elog(ERROR, "incomplete update info.");
if (oldtuple != NULL)
- store_change(ctx, CHANGE_UPDATE_OLD, oldtuple);
+ store_change(ctx, relation, CHANGE_UPDATE_OLD, oldtuple);
- store_change(ctx, CHANGE_UPDATE_NEW, newtuple);
+ store_change(ctx, relation, CHANGE_UPDATE_NEW, newtuple);
}
break;
case REORDER_BUFFER_CHANGE_DELETE:
@@ -146,7 +145,7 @@ plugin_change(LogicalDecodingContext *ctx, ReorderBufferTXN *txn,
if (oldtuple == NULL)
elog(ERROR, "incomplete delete info.");
- store_change(ctx, CHANGE_DELETE, oldtuple);
+ store_change(ctx, relation, CHANGE_DELETE, oldtuple);
}
break;
default:
@@ -164,39 +163,94 @@ plugin_change(LogicalDecodingContext *ctx, ReorderBufferTXN *txn,
/* Store concurrent data change. */
static void
-store_change(LogicalDecodingContext *ctx, ConcurrentChangeKind kind,
- HeapTuple tuple)
+store_change(LogicalDecodingContext *ctx, Relation relation,
+ ConcurrentChangeKind kind, HeapTuple tuple)
{
RepackDecodingState *dstate;
+ BufFile *file;
char kind_byte = (char) kind;
- bool flattened = false;
+ List *attrs_ext = NIL;
+ uint32 natt_ext;
dstate = (RepackDecodingState *) ctx->output_writer_private;
+ file = dstate->file;
/* Store the change kind. */
- BufFileWrite(dstate->file, &kind_byte, 1);
+ BufFileWrite(file, &kind_byte, 1);
/*
- * ReorderBufferCommit() stores the TOAST chunks in its private memory
- * context and frees them after having called apply_change(). Therefore
- * we need flat copy (including TOAST) that we eventually copy into the
- * memory context which is available to decode_concurrent_changes().
+ * If the tuple contains "external indirect" attributes, we need to write
+ * the contents to the file because we have no control over that memory.
*/
if (HeapTupleHasExternal(tuple))
{
- /*
- * toast_flatten_tuple_to_datum() might be more convenient but we
- * don't want the decompression it does.
- */
- tuple = toast_flatten_tuple(tuple, dstate->tupdesc);
- flattened = true;
+ TupleDesc desc;
+ Datum *attrs;
+ bool *isnull;
+
+ desc = RelationGetDescr(relation);
+ attrs = palloc0_array(Datum, desc->natts);
+ isnull = palloc0_array(bool, desc->natts);
+
+ heap_deform_tuple(tuple, desc, attrs, isnull);
+
+ /* First, gather and count the "external indirect" attributes. */
+ for (int i = 0; i < desc->natts; i++)
+ {
+ CompactAttribute *attr = TupleDescCompactAttr(desc, i);
+ varlena *varlena_pointer;
+
+ if (attr->attisdropped)
+ continue;
+
+ /* not a varlena datatype */
+ if (attr->attlen != -1)
+ continue;
+
+ /* no data */
+ if (isnull[i])
+ continue;
+
+ /* ok, we know we have a toast datum */
+ varlena_pointer = (varlena *) DatumGetPointer(attrs[i]);
+
+ if (!VARATT_IS_EXTERNAL(varlena_pointer))
+ continue;
+
+ if (VARATT_IS_EXTERNAL_INDIRECT(varlena_pointer))
+ attrs_ext = lappend(attrs_ext, varlena_pointer);
+ else
+ {
+ /*
+ * Logical decoding should not produce "external expanded"
+ * attributes (those actually should never appear on disk), so
+ * only TOASTed attribute can be seen here.
+ */
+ Assert(VARATT_IS_EXTERNAL_ONDISK(varlena_pointer));
+ }
+ }
+ natt_ext = list_length(attrs_ext);
}
- /* Store the tuple size ... */
- BufFileWrite(dstate->file, &tuple->t_len, sizeof(tuple->t_len));
- /* ... and the tuple itself. */
- BufFileWrite(dstate->file, tuple->t_data, tuple->t_len);
+ else
+ natt_ext = 0;
- /* Free the flat copy if created above. */
- if (flattened)
- pfree(tuple);
+ /* Write the number of external attributes. */
+ BufFileWrite(file, &natt_ext, sizeof(natt_ext));
+ /* ... and the attributes themselves, if there are some. */
+ foreach_ptr(varlena, attr_val, attrs_ext)
+ {
+ varlena *ext_val;
+ Size ext_val_size;
+
+ ext_val = detoast_external_attr(attr_val);
+ ext_val_size = VARSIZE_ANY(ext_val);
+ BufFileWrite(file, ext_val, ext_val_size);
+ pfree(ext_val);
+ }
+ list_free(attrs_ext);
+
+ /* Finally write the tuple size ... */
+ BufFileWrite(file, &tuple->t_len, sizeof(tuple->t_len));
+ /* ... and the tuple itself. */
+ BufFileWrite(file, tuple->t_data, tuple->t_len);
}
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index 1b05d5d418b..a32813e8530 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -65,11 +65,10 @@ typedef enum
*/
typedef struct RepackDecodingState
{
+#ifdef USE_ASSERT_CHECKING
/* The relation whose changes we're decoding. */
Oid relid;
-
- /* Tuple descriptor of the relation being processed. */
- TupleDesc tupdesc;
+#endif
/* The current output file. */
BufFile *file;
diff --git a/src/test/modules/injection_points/expected/repack_toast.out b/src/test/modules/injection_points/expected/repack_toast.out
index 4f866a74e32..b56dde134f8 100644
--- a/src/test/modules/injection_points/expected/repack_toast.out
+++ b/src/test/modules/injection_points/expected/repack_toast.out
@@ -13,6 +13,7 @@ step change:
UPDATE repack_test SET j=get_long_string() where i=2;
DELETE FROM repack_test WHERE i=3;
INSERT INTO repack_test(i, j) VALUES (4, get_long_string());
+ UPDATE repack_test SET i=3 where i=1;
step check2:
INSERT INTO relfilenodes(node)
diff --git a/src/test/modules/injection_points/specs/repack_toast.spec b/src/test/modules/injection_points/specs/repack_toast.spec
index b48abf21450..f84917daca2 100644
--- a/src/test/modules/injection_points/specs/repack_toast.spec
+++ b/src/test/modules/injection_points/specs/repack_toast.spec
@@ -74,10 +74,17 @@ teardown
session s2
step change
+# Separately test UPDATE of both plain ("i") and TOASTed ("j") attribute. In
+# the first case, the new tuple we get from reorderbuffer.c contains "j" as a
+# TOAST pointer, wich we need to update so it points to the new heap. In the
+# latter case, we receive "j" as "external indirect" value - here we test that
+# the decoding worker writes the tuple to a file correctly and that the
+# backend executing REPACK manages to restore it.
{
UPDATE repack_test SET j=get_long_string() where i=2;
DELETE FROM repack_test WHERE i=3;
INSERT INTO repack_test(i, j) VALUES (4, get_long_string());
+ UPDATE repack_test SET i=3 where i=1;
}
# Check the table from the perspective of s2.
step check2
--
2.47.3
--=-=-=--
^ permalink raw reply [nested|flat] 425+ messages in thread
end of thread, other threads:[~2026-02-27 18:01 UTC | newest]
Thread overview: 425+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2024-10-16 18:24 [PATCH v1 4/5] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-02-20 20:12 [PATCH v2 5/6] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-04-06 14:47 [PATCH v4 6/8] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2025-06-17 12:16 [PATCH v5 07/10] Allow to resize shared memory without restart Dmitrii Dolgov <9erthalion6@gmail.com>
2026-02-27 18:01 [PATCH 5/5] Serialize decoded tuples without flattening. Antonin Houska <ah@cybertec.at>
This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox