Обсуждение: 20.5.1
The following documentation comment has been logged on the website: Page: https://www.postgresql.org/docs/16/runtime-config-wal.html Description: The sentence in commit_siblings "A larger value makes it more probable that at least one other transaction will become ready to commit during the delay interval." seems to belong in commit_delay instead.
On Wednesday, February 7, 2024, PG Doc comments form <noreply@postgresql.org> wrote:
The following documentation comment has been logged on the website:
Page: https://www.postgresql.org/docs/16/runtime-config-wal. html
Description:
The sentence in commit_siblings "A larger value makes it more probable that
at least one other transaction will become ready to commit during the delay
interval." seems to belong in commit_delay instead.
That sentence in that location is correct. See “birthday paradox”.
Maybe phrasing it in the negative will make things clear - the more required sessions needed before allowing a delay the less likely the delay will be wasted.
David J.
On Wed, 2024-02-07 at 09:59 -0700, David G. Johnston wrote: > On Wednesday, February 7, 2024, PG Doc comments form <noreply@postgresql.org> wrote: > > The following documentation comment has been logged on the website: > > > > Page: https://www.postgresql.org/docs/16/runtime-config-wal.html > > Description: > > > > The sentence in commit_siblings "A larger value makes it more probable that > > at least one other transaction will become ready to commit during the delay > > interval." seems to belong in commit_delay instead. > > > > That sentence in that location is correct. See “birthday paradox”. To be more precise: if 15 other transactions are currently running, there is a bigger chance that at least one of them will want to flush WAL before "commit_delay" has expired than if there are only 3 other transactions. Yours, Laurenz Albe