On Mon, Mar 06, 2006 at 11:57:53AM +1100, Noel Faux wrote:
> >Anyway, if the block size is 8192
> >then 902292 sould be in the .6 file. If you can spare the time
> >then you might run the dd and od commands that Tom Lane mentions
> >in the above message and post the output.
> Here's the output:
I just noticed that the error in your original message was:
> pg_dump: Error message from server: ERROR: invalid page header in block
> 9022921 of relation "gap"
That's a different block than the one you said you found:
> Now after doing some searches I managed to work out that the data
> corruption starts at 902292.137 using this sql:
> SELECT * FROM gap WHERE ctid = '(902292,$x)'
> Where $x I changed from 1-150.
I suggested looking in the .6 file based on block 902292; if the
real bad block is 9022921 then I think it would be block 110025
in file .68 (again, check the math yourself).
--
Michael Fuhr