MIME-Version: 1.0
From: Bogdan Siara <bogdan.siara@gmail.com>
Date: Wed, 7 May 2025 13:00:18 +0200
Message-ID: <CAJCjmtsEtKMyBxZDg3Do72fr+VVF0eVUdt1aBUctD9BVEt_LNg@mail.gmail.com>
Subject: huge_pages=on cause could not map anonymous shared memory: Cannot
 allocate memory
To: pgsql-general@lists.postgresql.org
Content-Type: multipart/alternative; boundary="000000000000065609063489a5c1"
Archived-At: <https://www.postgresql.org/message-id/CAJCjmtsEtKMyBxZDg3Do72fr%2BVVF0eVUdt1aBUctD9BVEt_LNg%40mail.gmail.com>
Precedence: bulk

--000000000000065609063489a5c1
Content-Type: text/plain; charset="UTF-8"

Hi,
I have problem to run postgresql on ubuntu 24.04 server with huge_pages =
on. My instance have 8GB ram and 2 vcpus (t3a.large). My configuration is:

max_connections = 1000
superuser_reserved_connections = 3
shared_buffers = 1960MB
effective_cache_size = 5881MB
huge_pages = on
temp_buffers = 32MB
max_prepared_transactions = 100
work_mem = 1MB
maintenance_work_mem = 392MB
max_stack_depth = 4MB
dynamic_shared_memory_type = posix
archive_mode = on
max_wal_senders = 10
wal_sender_timeout = 0
min_wal_size = 4GB
max_wal_size = 16GB
wal_level = logical
checkpoint_timeout = 15min
checkpoint_completion_target = 0.9

In my sysct.conf I have:

vm.overcommit_memory=2
vm.overcommit_ratio=50
vm.vfs_cache_pressure=50
vm.dirty_background_ratio=10
vm.dirty_ratio=40
fs.nr_open=10000000
fs.file-max=10000000
vm.nr_hugepages=980
vm.hugetlb_shm_group=1010

my huge meminfo looks:

cat /proc/meminfo | grep -i huge
AnonHugePages:         0 kB
ShmemHugePages:        0 kB
FileHugePages:         0 kB
HugePages_Total:     980
HugePages_Free:      980
HugePages_Rsvd:        0
HugePages_Surp:        0
Hugepagesize:       2048 kB
Hugetlb:         2007040 kB

When I start postgres instance I get error:
May 07 10:56:32 pg1 pg_ctl[16753]: server starting
May 07 10:56:32 pg1 systemd[1]: Started postgres-17.4-local.service -
PostgreSQL 17.4 database local server.
May 07 10:56:35 pg1 pg_ctl[16756]: 2025-05-07 10:56:35.826 CEST [16756] - [
-  - ] [] : XX000FATAL:  could not map anonymous shared memory: Cannot
allocate memory
May 07 10:56:35 pg1 pg_ctl[16756]: 2025-05-07 10:56:35.826 CEST [16756] - [
-  - ] [] : XX000HINT:  This error usually means that PostgreSQL's request
for a shared memory segment exceeded available memory, swap space, or huge
pages. To reduce the request size (currently 2204106752 bytes), reduce
PostgreSQL's shared memory usage, perhaps by reducing "shared_buffers"
or "max_connections".
May 07 10:56:35 pg1 pg_ctl[16756]: 2025-05-07 10:56:35.826 CEST [16756] - [
-  - ] [] : 00000LOG:  database system is shut down
May 07 10:56:35 pg1 systemd[1]: postgres-17.4-sdx.service: Main process
exited, code=exited, status=1/FAILURE
May 07 10:56:35 pg1 systemd[1]: postgres-17.4-sdx.service: Failed with
result 'exit-code'.
May 07 10:56:35 pg1 systemd[1]: postgres-17.4-sdx.service: Consumed 3.493s
CPU time, 2.8M memory peak, 0B memory swap peak.

When I comment #huge_pages=on in configuration postgresql started without
problems.
Can someone tell me where is the problem?

Regards
Bogdan

--000000000000065609063489a5c1
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Hi,<div>I have problem to run postgresql on ubuntu 24.04 s=
erver with huge_pages =3D on. My instance have 8GB ram and 2 vcpus (t3a.lar=
ge). My configuration is:</div><div><br></div><div>max_connections =3D 1000=
<br>superuser_reserved_connections =3D 3<br>shared_buffers =3D 1960MB<br>ef=
fective_cache_size =3D 5881MB<br>huge_pages =3D on<br>temp_buffers =3D 32MB=
<br>max_prepared_transactions =3D 100<br>work_mem =3D 1MB<br>maintenance_wo=
rk_mem =3D 392MB<br>max_stack_depth =3D 4MB<br>dynamic_shared_memory_type =
=3D posix</div><div>archive_mode =3D on<br>max_wal_senders =3D 10<br>wal_se=
nder_timeout =3D 0<br>min_wal_size =3D 4GB <br>max_wal_size =3D 16GB<br>wal=
_level =3D logical<br>checkpoint_timeout =3D 15min<br>checkpoint_completion=
_target =3D 0.9</div><div><br></div><div>In my sysct.conf I have:</div><div=
><br></div><div>vm.overcommit_memory=3D2<br>vm.overcommit_ratio=3D50<br>vm.=
vfs_cache_pressure=3D50<br>vm.dirty_background_ratio=3D10<br>vm.dirty_ratio=
=3D40<br>fs.nr_open=3D10000000<br>fs.file-max=3D10000000<br>vm.nr_hugepages=
=3D980<br>vm.hugetlb_shm_group=3D1010</div><div><br></div><div>my huge memi=
nfo looks:<br><br>cat /proc/meminfo | grep -i huge<br>AnonHugePages: =C2=A0=
 =C2=A0 =C2=A0 =C2=A0 0 kB<br>ShmemHugePages: =C2=A0 =C2=A0 =C2=A0 =C2=A00 =
kB<br>FileHugePages: =C2=A0 =C2=A0 =C2=A0 =C2=A0 0 kB<br>HugePages_Total: =
=C2=A0 =C2=A0 980<br>HugePages_Free: =C2=A0 =C2=A0 =C2=A0980<br>HugePages_R=
svd: =C2=A0 =C2=A0 =C2=A0 =C2=A00<br>HugePages_Surp: =C2=A0 =C2=A0 =C2=A0 =
=C2=A00<br>Hugepagesize: =C2=A0 =C2=A0 =C2=A0 2048 kB<br>Hugetlb: =C2=A0 =
=C2=A0 =C2=A0 =C2=A0 2007040 kB</div><div><br></div><div>When I start postg=
res instance I get error:</div><div>May 07 10:56:32 pg1 pg_ctl[16753]: serv=
er starting<br>May 07 10:56:32 pg1 systemd[1]: Started postgres-17.4-local.=
service - PostgreSQL 17.4 database local server.<br>May 07 10:56:35 pg1 pg_=
ctl[16756]: 2025-05-07 10:56:35.826 CEST [16756] - [ - =C2=A0- ] [] : XX000=
FATAL: =C2=A0could not map anonymous shared memory: Cannot allocate memory<=
br>May 07 10:56:35 pg1 pg_ctl[16756]: 2025-05-07 10:56:35.826 CEST [16756] =
- [ - =C2=A0- ] [] : XX000HINT: =C2=A0This error usually means that Postgre=
SQL&#39;s request for a shared memory segment exceeded available memory, sw=
ap space, or huge pages. To reduce the request size (currently 2204106752 b=
ytes), reduce PostgreSQL&#39;s shared memory usage, perhaps by reducing &qu=
ot;shared_buffers&quot; or=C2=A0&quot;max_connections&quot;.<br>May 07 10:5=
6:35 pg1 pg_ctl[16756]: 2025-05-07 10:56:35.826 CEST [16756] - [ - =C2=A0- =
] [] : 00000LOG: =C2=A0database system is shut down<br>May 07 10:56:35 pg1 =
systemd[1]: postgres-17.4-sdx.service: Main process exited, code=3Dexited, =
status=3D1/FAILURE<br>May 07 10:56:35 pg1 systemd[1]: postgres-17.4-sdx.ser=
vice: Failed with result &#39;exit-code&#39;.<br>May 07 10:56:35 pg1 system=
d[1]: postgres-17.4-sdx.service: Consumed 3.493s CPU time, 2.8M memory peak=
, 0B memory swap peak.</div><div><br></div><div>When I comment #huge_pages=
=3Don in configuration postgresql started without problems.</div><div>Can s=
omeone tell me where is the problem?</div><div><br></div><div>Regards</div>=
<div>Bogdan</div></div>

--000000000000065609063489a5c1--