Re: Version 7.2.3 Vacuum abnormality

Andrew Sullivan <andrew@libertyrms.info> writes:
> On Thu, Feb 13, 2003 at 03:27:01PM +1000, Paul L Daniels wrote:
>> NOTICE:  Rel xamefiles: Uninitialized page 708135 - fixing
>> NOTICE:  Rel xamefiles: Uninitialized page 708136 - fixing
>> NOTICE:  Rel xamefiles: Uninitialized page 708137 - fixing

> This is a known and, it turns out, not real serious bug.  See, e.g., 
> http://archives.postgresql.org/pgsql-hackers/2002-11/msg00486.php
> I seem to recall Tom Lane noticing (uh, a few weeks ago, I think)
> that the problem happens not just on system tables.

The mechanism I described in the above-referenced message only occurs
for nailed-in-cache system tables.  Given Daniels' report (and one or
two others) I am suspicious that there's some path whereby rd_targblock
can fail to get reset after a vacuum for non-system tables too --- but
it hasn't been identified yet.

If what Daniels saw is due to a problem like that, then it's pretty
harmless.  If it's something else, the implications might be more dire.
        regards, tom lane