Re: catalog corruption causes - Mailing list pgsql-admin

From Kevin Grittner
Subject Re: catalog corruption causes
Date
Msg-id 4C5C14CE02000025000342D6@gw.wicourts.gov
Whole thread Raw
In response to catalog corruption causes  ("Burgholzer, Robert (DEQ)" <Robert.Burgholzer@deq.virginia.gov>)
Responses Re: catalog corruption causes  ("Burgholzer, Robert (DEQ)" <Robert.Burgholzer@deq.virginia.gov>)
List pgsql-admin
"Burgholzer, Robert (DEQ)" <Robert.Burgholzer@deq.virginia.gov>
wrote:

> I am trying to get my head around what causes catalog corruption.

From what I've seen many are caused by things outside of PostgreSQL
-- like bad disk, bad drivers, OS bugs, running on network drives
which aren't reliable, write caches without battery back-up,
anti-virus software, etc.

Another common cause is using faster-but-unreliable PostgreSQL
configuration settings like fsync=off or full_page_writes=off.  If
that's couple with a kill -9 or a crash of hardware or OS, you can
get corruption.

Then there's the possibility of PostgreSQL bugs.  Make sure you use
a recent minor release of whatever major release you're on, so you
get the benefit of bug fixes.  And I recommend staying away from
VACUUM FULL -- among the many other reasons there are to avoid it,
it seems to have more than it's share of odd corner cases where
things can go wrong.

Less likely, but still possible, is that there's some malicious
element involved.  Keep your security tight to minimize the risk of
that.

-Kevin

pgsql-admin by date:

Previous
From: "Burgholzer, Robert (DEQ)"
Date:
Subject: catalog corruption causes
Next
From: "Burgholzer, Robert (DEQ)"
Date:
Subject: Re: catalog corruption causes