From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | "Henrik Steffen" <steffen(at)city-map(dot)de> |
Cc: | "pg" <pgsql-general(at)postgresql(dot)org> |
Subject: | Re: Serious Crash last Friday |
Date: | 2002-07-11 06:36:14 |
Message-ID: | 14559.1026369374@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
"Henrik Steffen" <steffen(at)city-map(dot)de> writes:
> I now did a "pg_filedump -R 1048595 16584" on the file
> /var/lib/pgsql/data/base/16556/16584
What file is that? Doesn't look like it is an index ...
> It delivered the following output. Can you read anything from it?
Not a lot. I'd suggest "pg_filedump -i FILENAME" where FILENAME is the
OID of the toast-table index for your problem table. (Look at
pg_class.reltoastidxid if you're not sure.) That should produce a ton
of output along the lines of
Item 155 -- Length: 12 Offset: 4720 (0x1270) Flags: USED
Block Id: 4 linp Index: 39 Size: 12
Has Nulls: 0 Has Varlenas: 0
What you want to look for is outrageously large values in the "Block Id"
field. Once you find 'em, a "pg_filedump -i -f" of just the block
containing the broken item(s) would be worth studying.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Henrik Steffen | 2002-07-11 06:37:20 | Re: Serious Crash last Friday |
Previous Message | Curt Sampson | 2002-07-11 06:35:59 | Re: Linux max on shared buffers? |