Do you have any other suggestions as to which tables generally can be excluded from the autovacuum based on the usage patterns? Can you see anything with respect to my new autovacuum parameters that may cause issue's and are there any other parameters that you suggest I need to change to make the autovacuum daemon more aggressive? PS: Currently we have the Cluster command running on the sipaccounts table as the vacuum full is taking too long. It would be nice though to have some piece of mind that the cluster command is mvcc safe, as Heikki and Aidan have mentioned that it is not and may break things in our environment, I'm a little afraid of running with the cluster command, and should possibly go back to the vacuum full :/ Thanks all for any and all suggestions/comments. Thanks Bruce "Tom Lane" <tgl@xxxxxxxxxxxxx> wrote in message news:19106.1173111068@xxxxxxxxxxxxxxxx > "Bruce McAlister" <bruce.mcalister@xxxxxxxxxxx> writes: >> [1] AutoVacuum runs during the day over the entire PostgreSQL cluster, > > Good, but evidently you need to make it more aggressive. > >> [2] A Vacuum Full Verbose is run during our least busy period (generally >> 03:30) against the Database, > >> [3] A Re-Index on the table is performed, > >> [4] A Cluster on the table is performed against the most used index, > >> [5] A Vacuum Analyze Verbose is run against the database. > > That is enormous overkill. Steps 2 and 3 are a 100% waste of time if > you are going to cluster in step 4. Just do the CLUSTER and then > ANALYZE (or VACUUM ANALYZE if you really must, but the value is marginal). > > regards, tom lane > > ---------------------------(end of broadcast)--------------------------- > TIP 3: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faq >
