Re: Vacuum deadlocks?
От | eric soroos |
---|---|
Тема | Re: Vacuum deadlocks? |
Дата | |
Msg-id | 124755539.1169930075@[4.42.179.151] обсуждение исходный текст |
Ответ на | Re: Vacuum deadlocks? ("Josh Berkus" <josh@agliodbs.com>) |
Ответы |
Re: Vacuum deadlocks?
|
Список | pgsql-novice |
Thanks Josh. > I think there were some issues with VACUUM on 7.2.1. Were I you, the > first thing I would try is an upgrade to 7.2.3, which will also fix > some serious security issues and at least one backup/restore issue. > Maybe someone else on the list will speak to your specific issue, or > you can look at the release notes for 7.2.2 and 7.2.3. Great. The packager's site is down. (I knew there was a reason for that nagging feeling that I should be building from sourceinstead of using a package. ) > Second, were you aware that you have the option of running a regular > VACUUM or VACUUM <table-name> from a database connection without > locking users out of the database? In high-transaction environments, > I can run these every 5 -15 minutes. The vacuumdb commmand-line > utility, I believe, does a VACUUM FULL which exclusively locks the > database (and you do need to do periodically, just not frequently). I looked at the docs and didn't see the mention that vacuumdb did a vacuum full. That would certainly explain the deadlocking. But it seems that that is not the case. from the 7.2.1 docs: -f Perform "full" vacuuming. -a Vacuum all databases. .. vacuumdb is a shell script wrapper around the backend command VACUUM via the PostgreSQL interactive terminal psql. Thereis no effective difference between vacuuming databases via this or other methods. eric
В списке pgsql-novice по дате отправления: