Re: Help tracking down problem with inserts slowing

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: swampler(at)noao(dot)edu
Cc: Robert Treat <xzilla(at)users(dot)sourceforge(dot)net>, Postgres-performance <pgsql-performance(at)postgresql(dot)org>
Subject: Re: Help tracking down problem with inserts slowing
Date: 2003-12-08 15:35:49
Message-ID: 9618.1070897749@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

Steve Wampler <swampler(at)noao(dot)edu> writes:
> Thanks, Tom. Are there any reasons why it would not appear?:

Oh, I shoulda read the code more carefully. I was looking at the bottom
of lazy_scan_index, where the printout is done, and failed to notice the
test at the top:

/*
* If the index is not partial, skip the scan, and just assume it has
* the same number of tuples as the heap.
*/

So for ordinary indexes, nothing will appear unless vacuum has actual
work to do (that is, it recycled at least one dead tuple in the table).

Short answer: update or delete some row in the table, and then try
vacuum verbose.

Alternatively, you can just look at the pg_class row for the index.
relpages and reltuples will contain the info you are after ... and
they are certainly up to date at this point ;-)

regards, tom lane

In response to

Browse pgsql-performance by date

  From Date Subject
Next Message Jack Coates 2003-12-08 17:43:45 Re: tuning questions
Previous Message Steve Wampler 2003-12-08 15:14:53 Re: Help tracking down problem with inserts slowing