Re: PostgreSQL 8.1.0 catalog corruption

On Nov 21, 2005, at 5:50 PM, Tom Lane wrote:

> Bob Ippolito <bob@redivi.com> writes:
>> I don't touch pg_class at all... this is what I'm doing (over and
>> over again).
>
>>     -- clone_table is almost always a no-op, but once a day it creates a
>> new table
>>          SELECT clone_table('ping', 'ping_%s', '')
>>          SELECT drop_ping_constraints('ping_%s')
>>     -- stuff that doesn't effect DDL
>>     SELECT add_ping_constraints('ping_%s')
>
> Hm, do the drop/add constraint functions get executed even when
> clone_table decides not to make a new table?  If so, that would  
> probably
> explain the pattern I'm seeing in the dump of many updates of the
> pg_class row.

Yes, they do.  The constraints are there for constraint exclusion.

> This still doesn't give us a hint why the row disappeared, but  
> maybe we
> can try running these functions for awhile and see if anyone can
> reproduce a failure.

If it matters, I have had the same code running on Bizgres 0.7.4 for  
quite some time with no issues at all.  I may just have to migrate  
the test server to Bizgres 0.8 if we can't figure out why PostgreSQL  
8.1.0 choked here.

-bob