PostgreSQL 8.4.4 on x86_64-pc-solaris2.10, compiled by GCC gcc (GCC) 3.4.3 (csl-sol210-3_4-branch+sol_rpath), 64-bit
2 large tables (around a billion rows each) recently received attention from autovacuum. They both currently show up
inmy currently running queries as "autovacuum: VACUUM tablename (to prevent wraparound)".
These tables are insert-only audit tables….no updates, no deletes.
Autovacuum settings are default with no overrides for these specific tables (the only override is to set 10
max_workers)
I will admit that the "Preventing Transaction ID Wraparound Failures" chapter in the manual (and the subsequent
autovacuumdaemon chapter) make my head spin.
Is there some specific query related to relfrozenxid, datfrozenxid that I should be utilizing to do some analysis?
ObviouslySOME threshhold has been reached…hence autovacuum stepping in.
My two real questions are:
1. Am I in any real danger here? I have run autovacuum continuously since the dump-restore that created this
database6 months ago.
2. These autovacuums have been running for 5 days now. I have a database maintenance window tonight where I have to
takedown the server. I assume autovacuum will re-awaken and decide to start vacuuming these again. I assume I am
effectingperformance with these needless? vacuums. Any tweak I can/should make to any of the freeze_age params to
preventthis?