Re: 8.2.3 PANIC with "corrupted item pointer"

"Henka" <henka(at)cityweb(dot)co(dot)za> writes:

>> Other than that it might be interesting to know the values of some server
>> parameters: "fsync" and "full_page_writes". Have you ever had this machine
>> crash or had a power failure? And what kind of i/o controller is this?
>
> fsync = off
> full_page_writes = default
>
> Sadly yes, the machine has experienced a power failure about 3 weeks ago
> (genset startup problem). With fsync=off this presents a problem wrt safe
> recovery, I know...

Ugh. The worst part is that you won't even know that there's anything wrong
with your data. I would actually suggest that if you run with fsync off and
have a power failure or kernel crash you should just immediately restore from
your last backup and not risk running with the possibly corrupt database.

Honestly this seems like a weird error to occur as a result of crashing with
fsync off but anything's possible. More likely is you have records that you
have partial transactions in your database, ie, records which were inserted or
deleted in a transaction but missing other records that were inserted or
deleted in the same transaction.

You could probably fix this particular problem by reindexing the corrupted
index. But you may never know if some of the data is incorrect.

--
Gregory Stark
EnterpriseDB http://www.enterprisedb.com