public inbox for [email protected]help / color / mirror / Atom feed
Wal_keep_size 6+ messages / 3 participants [nested] [flat]
* Wal_keep_size @ 2025-10-06 19:50 Raj <[email protected]> 0 siblings, 1 reply; 6+ messages in thread From: Raj @ 2025-10-06 19:50 UTC (permalink / raw) To: Pgsql-admin <[email protected]> Should Wal_keep_size <= max_wal_size ? ^ permalink raw reply [nested|flat] 6+ messages in thread
* Re: Wal_keep_size @ 2025-10-06 21:38 Thomas Carroll <[email protected]> parent: Raj <[email protected]> 0 siblings, 1 reply; 6+ messages in thread From: Thomas Carroll @ 2025-10-06 21:38 UTC (permalink / raw) To: Pgsql-admin <[email protected]>; Raj <[email protected]> Hope nobody minds me chipping in here. The answer is "not necessarily." The names make them seem closely related, but not so much. max_wal_size is all about checkpoints. When the WAL exceeds this, Postgres will try to run a checkpoint. Under some circumstances it won't, but let's keep it simple for now. wal_keep_size is all about replication. Replicas need WAL files to stick around for long enough that they can process them. wal_keep_size is a minimum figure. Tom On Monday, October 6, 2025 at 03:51:17 PM EDT, Raj <[email protected]> wrote: Should Wal_keep_size <= max_wal_size ? ^ permalink raw reply [nested|flat] 6+ messages in thread
* Re: Wal_keep_size @ 2025-10-07 04:52 Raj <[email protected]> parent: Thomas Carroll <[email protected]> 0 siblings, 2 replies; 6+ messages in thread From: Raj @ 2025-10-07 04:52 UTC (permalink / raw) To: Thomas Carroll <[email protected]>; +Cc: Pgsql-admin <[email protected]> If wal_keep_size is more than max_wal_size wouldn't it always trigger checkpoint ? On Tue, 7 Oct 2025, 03:08 Thomas Carroll, <[email protected]> wrote: > Hope nobody minds me chipping in here. > > The answer is "not necessarily." The names make them seem closely > related, but not so much. > > max_wal_size is all about *checkpoints*. When the WAL exceeds this, > Postgres will try to run a checkpoint. Under some circumstances it won't, > but let's keep it simple for now. > > wal_keep_size is all about *replication*. Replicas need WAL files to > stick around for long enough that they can process them. wal_keep_size is > a minimum figure. > > Tom > On Monday, October 6, 2025 at 03:51:17 PM EDT, Raj < > [email protected]> wrote: > > > Should Wal_keep_size <= max_wal_size ? > ^ permalink raw reply [nested|flat] 6+ messages in thread
* Re: Wal_keep_size @ 2025-10-07 04:54 Raj <[email protected]> parent: Raj <[email protected]> 1 sibling, 0 replies; 6+ messages in thread From: Raj @ 2025-10-07 04:54 UTC (permalink / raw) To: Thomas Carroll <[email protected]>; +Cc: Pgsql-admin <[email protected]> Note archive mode is enabled. On Tue, 7 Oct 2025, 10:22 Raj, <[email protected]> wrote: > If wal_keep_size is more than max_wal_size wouldn't it always trigger > checkpoint ? > > On Tue, 7 Oct 2025, 03:08 Thomas Carroll, <[email protected]> wrote: > >> Hope nobody minds me chipping in here. >> >> The answer is "not necessarily." The names make them seem closely >> related, but not so much. >> >> max_wal_size is all about *checkpoints*. When the WAL exceeds this, >> Postgres will try to run a checkpoint. Under some circumstances it won't, >> but let's keep it simple for now. >> >> wal_keep_size is all about *replication*. Replicas need WAL files to >> stick around for long enough that they can process them. wal_keep_size is >> a minimum figure. >> >> Tom >> On Monday, October 6, 2025 at 03:51:17 PM EDT, Raj < >> [email protected]> wrote: >> >> >> Should Wal_keep_size <= max_wal_size ? >> > ^ permalink raw reply [nested|flat] 6+ messages in thread
* Re: Wal_keep_size @ 2025-10-07 05:28 Laurenz Albe <[email protected]> parent: Raj <[email protected]> 1 sibling, 1 reply; 6+ messages in thread From: Laurenz Albe @ 2025-10-07 05:28 UTC (permalink / raw) To: Raj <[email protected]>; Thomas Carroll <[email protected]>; +Cc: Pgsql-admin <[email protected]> On Tue, 2025-10-07 at 10:22 +0530, Raj wrote: > If wal_keep_size is more than max_wal_size wouldn't it always trigger checkpoint ? No. "max_wal_size" has no direct connection with the size of WAL (and I think we should have chosen a different name for that parameter). The parameter is about the amount of WAL *generated since the latest checkpoint*, not about the absolute amount of WAL present. Yours, Laurenz Albe ^ permalink raw reply [nested|flat] 6+ messages in thread
* Re: Wal_keep_size @ 2025-10-12 07:55 Raj <[email protected]> parent: Laurenz Albe <[email protected]> 0 siblings, 0 replies; 6+ messages in thread From: Raj @ 2025-10-12 07:55 UTC (permalink / raw) To: Laurenz Albe <[email protected]>; +Cc: Thomas Carroll <[email protected]>; Pgsql-admin <[email protected]> Thanks for the clarification. On Tue, 7 Oct 2025 at 10:58, Laurenz Albe <[email protected]> wrote: > On Tue, 2025-10-07 at 10:22 +0530, Raj wrote: > > If wal_keep_size is more than max_wal_size wouldn't it always trigger > checkpoint ? > > No. "max_wal_size" has no direct connection with the size of WAL (and I > think we should have chosen a different name for that parameter). The > parameter is about the amount of WAL *generated since the latest > checkpoint*, > not about the absolute amount of WAL present. > > Yours, > Laurenz Albe > ^ permalink raw reply [nested|flat] 6+ messages in thread
end of thread, other threads:[~2025-10-12 07:55 UTC | newest] Thread overview: 6+ messages (download: mbox.gz follow: Atom feed) -- links below jump to the message on this page -- 2025-10-06 19:50 Wal_keep_size Raj <[email protected]> 2025-10-06 21:38 ` Thomas Carroll <[email protected]> 2025-10-07 04:52 ` Raj <[email protected]> 2025-10-07 04:54 ` Raj <[email protected]> 2025-10-07 05:28 ` Laurenz Albe <[email protected]> 2025-10-12 07:55 ` Raj <[email protected]>
This inbox is served by agora; see mirroring instructions for how to clone and mirror all data and code used for this inbox