Re: [HACKERS] Re: vacuum... - Mailing list pgsql-hackers

From Vadim B. Mikheev
Subject Re: [HACKERS] Re: vacuum...
Date
Msg-id 3508A973.7B28D40B@sable.krasnoyarsk.su
Whole thread Raw
In response to Re: vacuum...  (The Hermit Hacker <scrappy@hub.org>)
List pgsql-hackers
The Hermit Hacker wrote:
>
>         psql -c "vacuum analyze $i" <database>
                          ^^^^^^^
It takes additional time and shouldn't be used in each run.

> >   I was curious about your thoughts on the problem
> > with the corrupted index "FATAL 1:btree BTP_CHAIN was expected".
> > Was this a known bug in pre-6.3 versions? We have experienced
> > spurious postgres process core dumps. Can this corrupt the
> > indexes/db?

Yes. Core dumps and power off-s can break indices in all versions.

> >   You suggested that we upgrade to 6.3 and I agree
> > that we will do this in the very near future. One
> > question regarding this. Is regular vacuuming still
> > a requirement in 6.3?
>
>         regular vacuuming will always be a requirement, as it updates
> various statistics...but, one of the things that I believe is being looked
> into is having the database overwrite delete records, so that a vacuum
> isn't required to remove those records and shrink the database...

Required, currently.

Vadim

pgsql-hackers by date:

Previous
From: The Hermit Hacker
Date:
Subject: Re: vacuum...
Next
From: Michael Graff
Date:
Subject: RVM -- recoverable virtual memory