Thread: Re: how to use pg_resetxlog - closer ...

Thanks to all ... I think I'm closer to at leasting
something useful as far as the database is concern.

Recap:  Running PostgreSQL 7.2.1 on RedHat Linux 7.2
and I just experienced a RAID failure with the DB
still active.  Tom pointed me in the area of the file
'1249' which was no longer in my database directory.
I copied that from anther directory and got part
of the way there.  There were some errors regarding
missing files in pg_clog - I faked that, too.

Now:  I am getting an error  when I try to do a select
from any table :

[snip]
cannot find attribute 1 of relation <table>
[/snip]

Looking in the groups.google.com, I see similar things
which was fixed with running PostgreSQL in stand alone
mode and forcing a reindex.  When I tried that, I got:

[snip error]

[/snip]
backend> reindex database testdb force ;
DEBUG:  StartTransactionCommand
DEBUG:  query: reindex database testdb force ;

DEBUG:  ProcessUtility: reindex database testdb force ;

ERROR:  Cannot create unique index. Table contains non-unique values
DEBUG:  AbortCurrentTransaction
[snip error]

Is there a way around this so I can at least dump some
data?

Thanks bunches!

-X

"Johnson, Shaunn" <SJohnson6@bcbsm.com> writes:
> Recap:  Running PostgreSQL 7.2.1 on RedHat Linux 7.2
> and I just experienced a RAID failure with the DB
> still active.  Tom pointed me in the area of the file
> '1249' which was no longer in my database directory.
> I copied that from anther directory and got part
> of the way there.

Uh, that will most definitely NOT work.  You've installed some other
database's list of table columns.  If you can't resurrect the original
pg_attribute file for this database (did you look in /lost+found?)
then I think you are well and truly screwed :-(.  Time to revert to
backup, because there is no hope of extracting anything useful.

            regards, tom lane