Re: Profiling - Mailing list pgsql-performance

From Tom Lane
Subject Re: Profiling
Date
Msg-id 6741.1040136556@sss.pgh.pa.us
Whole thread Raw
In response to Re: Profiling  (Cory 'G' Watson <gphat@cafes.net>)
List pgsql-performance
"Cory 'G' Watson" <gphat@cafes.net> writes:
> I assume I do not need a --analyze, since that table has no indexes.

Whether you need analyze or not has nothing to do with whether there
are indexes.  You probably don't need it once an hour, but maybe once
a day would be good.

> Should I vacuum the entire DB?

Overkill; just get the heavily-updated table(s).  A DB-wide vacuum must
be done occasionally, but again once-a-day would be plenty.

> Any other settings I should look at?

Free space map (fsm) settings must be adequate to keep track of the free
space in your tables.

However, all of this relates only to keeping performance good on the
table with lots of updates.  If you are seeing progressive degradation
on a table that only gets INSERTs, then there's something else going on.
AFAIR you didn't show us an EXPLAIN ANALYZE for the principal query?

            regards, tom lane

pgsql-performance by date:

Previous
From: "Shridhar Daithankar"
Date:
Subject: Re: Profiling
Next
From: "scott.marlowe"
Date:
Subject: Re: Profiling