Re: Vacuum Vs Vacuum Full - Mailing list pgsql-general

From Tom Lane
Subject Re: Vacuum Vs Vacuum Full
Date
Msg-id 10552.1217956535@sss.pgh.pa.us
Whole thread Raw
In response to Re: Vacuum Vs Vacuum Full  ("Matthew T. O'Connor" <matthew@zeut.net>)
List pgsql-general
"Matthew T. O'Connor" <matthew@zeut.net> writes:
> Tom Lane wrote:
>> 8.2.  But you could still get that message, even in CVS HEAD, if
>> autovacuum was failing to complete for some reason (and had been
>> failing for quite a long time).

> Should that message to updated since a database-wide vacuum is no longer
> required, or are you saying that the message is still relevant is some
> corner cases?

I think the message is okay as-is, or at least that the code doesn't
have the information available to do better --- it knows which database
is the most problematic, but not which table(s) within that DB most need
vacuuming.  So the easiest manual fix is still a DB-wide vacuum.
Besides which, if you've got one problem table then you've probably got
more than one.

The odds of anyone seeing this message in the field in 8.2 or later seem
pretty remote anyway, so I'm not feeling like we should expend
tremendous effort to make it better.

            regards, tom lane

pgsql-general by date:

Previous
From: "Matthew T. O'Connor"
Date:
Subject: Re: Vacuum Vs Vacuum Full
Next
From: "Kynn Jones"
Date:
Subject: max_fsm_relations question