From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | "Alok K(dot) Dhir" <alok(at)dhir(dot)net> |
Cc: | pgsql-general(at)postgresql(dot)org |
Subject: | Re: Please help recover data - lost pg_control |
Date: | 2001-07-03 15:58:56 |
Message-ID: | 18867.994175936@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
"Alok K. Dhir" <alok(at)dhir(dot)net> writes:
> ERROR: readDatum: ']' expected, length = 12
Hm, are you sure that was 7.1b3, and not something even older? Looking
at the CVS logs, the last change in stored rule format was between beta1
and beta2, so I'd expect this sort of failure if you tried to run 7.1
release against a 7.1b1 database (noting that you clobbered the normal
DB version crosscheck by running pg_resetxlog).
You might be able to build a hacked version of 7.1 release that will
read your 7.1b1 database by undoing the CONST node dump format change of
2001-01-07 --- see rev 1.103 at
http://www.ca.postgresql.org/cgi/cvsweb.cgi/pgsql/src/backend/nodes/readfuncs.c
but note that most of the diffs don't need to be undone, you just need
to change back the order of the fields that _readConst is expecting.
I'm not sure that's the only thing that'll bite you, but it's worth a
try. If you get a clean dump, initdb with a non-hacked copy of 7.1.2
before restoring.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Richard Huxton | 2001-07-03 16:12:43 | Re: Determining scan types |
Previous Message | Peter Eisentraut | 2001-07-03 15:46:15 | Re: Regression |