Thread: Re: [COMMITTERS] pgsql: Make large sequential scans and VACUUMs work in a limited-size
Re: [COMMITTERS] pgsql: Make large sequential scans and VACUUMs work in a limited-size
From
Alvaro Herrera
Date:
Tom Lane wrote: > Log Message: > ----------- > Make large sequential scans and VACUUMs work in a limited-size "ring" of > buffers, rather than blowing out the whole shared-buffer arena. Aside from > avoiding cache spoliation, this fixes the problem that VACUUM formerly tended > to cause a WAL flush for every page it modified, because we had it hacked to > use only a single buffer. Those flushes will now occur only once per > ring-ful. The exact ring size, and the threshold for seqscans to switch into > the ring usage pattern, remain under debate; but the infrastructure seems > done. The key bit of infrastructure is a new optional BufferAccessStrategy > object that can be passed to ReadBuffer operations; this replaces the former > StrategyHintVacuum API. I think now is time to re-test the patch for advancing OldestXmin during vacuum? -- Alvaro Herrera http://www.CommandPrompt.com/ The PostgreSQL Company - Command Prompt, Inc.
Re: [COMMITTERS] pgsql: Make large sequential scans and VACUUMs work in a limited-size
From
Heikki Linnakangas
Date:
Alvaro Herrera wrote: > Tom Lane wrote: >> Log Message: >> ----------- >> Make large sequential scans and VACUUMs work in a limited-size "ring" of >> buffers, rather than blowing out the whole shared-buffer arena. Aside from >> avoiding cache spoliation, this fixes the problem that VACUUM formerly tended >> to cause a WAL flush for every page it modified, because we had it hacked to >> use only a single buffer. Those flushes will now occur only once per >> ring-ful. The exact ring size, and the threshold for seqscans to switch into >> the ring usage pattern, remain under debate; but the infrastructure seems >> done. The key bit of infrastructure is a new optional BufferAccessStrategy >> object that can be passed to ReadBuffer operations; this replaces the former >> StrategyHintVacuum API. > > I think now is time to re-test the patch for advancing OldestXmin during > vacuum? Thanks for the reminder, I'll schedule those tests. -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com