Error in postgresql after crash unable to restart this needs to be posted on -hackers or -bugs, so that tom, oleg and teodor can take a look

anil maran wrote:

<AndrewSN> this needs to be posted on -hackers or -bugs, so that tom, oleg and teodor can take a look
Logs here
http://rafb.net/paste/results/NgHkIl17.html

Is this the same message you posted on -general? The error message was

PANIC: block 1100 unfound

I see several occurences of that message in gistxlog.c. Can you restart
the postmaster with "log_error_verbosity" set to "verbose" on
postgresql.conf, so that we have more details about it? Please post the
messages here on the list, not on a paste site.

The next problem is determining which index is having the problem, and
seeing if it has 1100 blocks or not. Maybe an extend operation was not
logged, or something.

--
Alvaro Herrera http://www.CommandPrompt.com/
PostgreSQL Replication, Consulting, Custom Development, 24x7 support

Alvaro Herrera <alvherre@commandprompt.com> writes:

anil maran wrote:
PANIC: block 1100 unfound

The next problem is determining which index is having the problem, and
seeing if it has 1100 blocks or not. Maybe an extend operation was not
logged, or something.

I'm thinking this might be the GIST variant of that
replay-after-truncation problem we fixed back in March. That is, the
xlog replay code isn't prepared for the possibility that the block it's
supposed to change no longer exists.

regards, tom lane