Re: Still big problems with pg_dump! - Mailing list pgsql-general

From Wim
Subject Re: Still big problems with pg_dump!
Date
Msg-id 3D873B69.1000105@belbone.be
Whole thread Raw
In response to Still big problems with pg_dump!  (Wim <wdh@belbone.be>)
Responses Re: Still big problems with pg_dump!  (Andrew Sullivan <andrew@libertyrms.info>)
List pgsql-general

Andrew Sullivan wrote:

>-hackers removed.
>
>On Tue, Sep 17, 2002 at 10:11:41AM +0200, Wim wrote:
>
>
>
>>ERROR:  AllocSetFree: cannot find block containing chunk 4c5ad0
>>
>>
>
>This is definitely some sort of disk problem.  Either you've written
>bad data to the disk in some way, or else the disk is corrupted or
>damaged.
>
>If it is a hardware problem, the obvious suspects are memory (I'd
>discount this idea unless everything else doesn't check out), a disk
>failure, or a controller failure.
>
>It could be OS related as well.  Several of the 2.4 Linux kernel
>series, for instance, had roblems with massive filesystem corruption.
>
>
Postgres is running on solaris 8...
It is the same database as previous time that has the problem, but not
the same table.
I think it's an error is the system tables.

>
>
>>Some people suggest a drive failure, but I checked that and found no
>>problems...
>>
>>
>
>How did you check?
>
>
with fsck.

>
>
>>I must say that one of the table contains more than 3.000.000 rows,
>>another more than 1.400.000...
>>
>>
>
>When is your most recent backup?  If you can't pg_dump, you will be
>needing that backup.
>
>
Have backup... I can still SQL COPY to a text file, so that's no problem
so far.

>
>
>>I must say that I had this problem a few months before, I got some help
>>then, but that couldn't solve my problem,
>>I recreated the database from scratch and copied the data, to fix thing
>>quickly. Thing went well for about two months :-(
>>
>>
>
>So you re-installed the data set on a machine that had somehow
>failed, you don't know why, and hoped that the problem would
>solve itself?  Uh, that wasn't a good idea.  In the future, if you
>have a problem which people suggest might be, for instance, a bad
>disk, it'd be a _very good_ idea to figure out precisely what the
>problem is before relying on the identical hardware again.
>
>A
>
>
>
I know, I don't have much spare hardware, and the database had to work
quickly, it was the
only solution then.
Checked the disk, reinstalled the OS and still waiting for a CPU and
memory upgrade.



pgsql-general by date:

Previous
From: Tom Lane
Date:
Subject: Re: [HACKERS] PostgreSQL 7.3: help on new CREATE TYPE
Next
From: "Mihai Gheorghiu"
Date:
Subject: Help!