Re: Uninitialized pages notices - Mailing list pgsql-admin

From Timothy D McKernan
Subject Re: Uninitialized pages notices
Date
Msg-id 3DCFD6AC.30401@dyrectmedia.com
Whole thread Raw
In response to Re: Vacuum return codes (vacuum as db integrity check?)  ("Stephane Charette" <stephanecharette@telus.net>)
List pgsql-admin
Tom Lane wrote:
> Timothy D McKernan <tdm4@dyrectmedia.com> writes:
>
>>Nov  6 04:17:09 dolidb-n1 logger: NOTICE:  Rel pg_type: Uninitialized
>>page 6 - fixing
>
>
>>I'd like to know what this means to our system - what causes an
>>uninitialized page?  What could cause it to be so infrequent?
>
>
> "Uninitialized" means "page contains zeroes" (or at least a few critical
> page-header fields contain zeroes, which they should never do).  Usually
> I take this as an indication of hardware problems.  But:
>
>
>>Here are our system details:
>>postgre: 7.2.3 (~4GB in size)
>>os: Redhat 7.1sbe (Seawolf)
>>kernel: 2.4.9-12smp
>>ram: ~1GB ECC
>
>
> SMP?  Are you actually using SMP hardware?  I seem to recall that the
> 2.4 Linux kernels weren't stable on SMP machines till 2.4.15 or
> thereabouts.  In any case, RedHat 7.1 is pretty long in the tooth.
> Perhaps an OS update would make life better.

Yes, we have a dual processor board.  Per your suggestion we'll be
upgrading the system soon and watching the results of that.

Thanks,
Tim


>
>             regards, tom lane
>


--
Dyrect Media Group
P.O. Box 486
6000 Goodrich Rd
Clarence Center, NY  14032-0486

OFFICE: 716-504-1141 ext 208
CELL: 716-510-2451
AIM: somecallmetim100


pgsql-admin by date:

Previous
From: Jodi Kanter
Date:
Subject: constraint notices
Next
From: Gary DeSorbo
Date:
Subject: Monitor Tools