Обсуждение: BUG #3567: invalid page header in block XXXXof relation

The following bug has been logged online:

Bug reference:      3567
Logged by:          sapna kapoor
Email address:      sapna.kapoor@nic.in
PostgreSQL version: 8.1.9
Operating system:   RHEL 5.0 AS
Description:        invalid page header in block XXXXof relation
Details:

We have database of 300 tables ,
some tables are giving error :
ERROR:  invalid page header in block 32655 of relation "AAAA" while
accessing.

Some are giving
ERROR:  could not access status of transaction 1778384896
DETAIL:  could not open file "pg_clog/06A0": No such file or directory

We are in the process of loading data from distibuted locations. We are
importing data onto this server from the export files taken from distributed
locations.

"sapna kapoor" <sapna.kapoor@nic.in> writes:
> some tables are giving error :
> ERROR:  invalid page header in block 32655 of relation "AAAA" while
> accessing.

> Some are giving
> ERROR:  could not access status of transaction 1778384896
> DETAIL:  could not open file "pg_clog/06A0": No such file or directory

If you had just one of these, I might think it suggested a software
problem.  But multiple spots of data corruption sound more like flaky
hardware.  I'd suggest running disk and memory tests ...

            regards, tom lane

Tom Lane wrote:
> "sapna kapoor" <sapna.kapoor@nic.in> writes:
>> some tables are giving error :
>> ERROR:  invalid page header in block 32655 of relation "AAAA" while
>> accessing.
>
>> Some are giving
>> ERROR:  could not access status of transaction 1778384896
>> DETAIL:  could not open file "pg_clog/06A0": No such file or directory
>
> If you had just one of these, I might think it suggested a software
> problem.

The similar clog problem was reported there
http://archives.postgresql.org/pgsql-bugs/2007-07/msg00124.php.

It looks like vacuum truncate clog file but, some tuples still have xmin or xmax
from truncated clog.

Sapna,

can
You try following http://pgfoundry.org/projects/pgcheck/ project. It is still in
alpha, but page check
(http://cvs.pgfoundry.org/cgi-bin/cvsweb.cgi/pgcheck/pokus/pgcheck_page/) should
be work.

        Zdenek