Re: Race-condition with failed block-write?

Arjen van der Meijden wrote:
> On 13-9-2005 23:01, Tom Lane wrote:
>
> I'll try and see if I can dig up more from the logs and see if I can
> test a few reasonable scenario's tomorrow though.

I wasn't able to find anything suspicious at all. A smart-shutdown
should've come from the user postgres and I can trace all the 'pg_ctl
...'-calls for that postgres-installation to other dates than Sep 1 or
the 8.1-installation.
Besides that, if the init.d-script with its fast shutdown wouldn't have
responded, it seems very unlikely to me that I'd try a normal shutdown
after the fast shutdown. I would've tried an immediate shutdown instead.

Besides that the settings hadn't been changed in days at that time, so
any query issued should've been logged (log statement was enabled), even
if it got cancelled before it finished. I also can't find kill's from
root or similar statements, pointing to other incorrect usages.

Nor was I silly enough to start postgres 8.1 on the 8.0.3-data-store, or
postgresql 8.0.3 with pg_ctl instead of the init.d-script.

So why it was able to do two shutdowns in a row, I have no clue, not
even how to think of a reasonable scenario what happened. I wasn't
actively using/testing the database at that time. The machine itself
wasn't shut down as well at that date.

Best regards,

Arjen