Razvan Surdulescu <surdules@yahoo.com> writes:
>>> cspan=> update contact set gender = 'M';
>>> ERROR: cache lookup failed for function 70529
>>> The database has a few GIST (full-search, tsearch2) indices, but
>>> otherwise is pretty ordinary.
>> Any of those on "contact"?
> Yes, one of them, see the contact schema below.
I recall some discussion recently about the FTI code getting confused
with this sort of symptom resulting; you might want to check the
archives.
>> What would be useful to look at is a debugger stack trace from the core
>> dump...
> Hmm, I cannot find a core file from the crash. I looked in the
> postgres/bin directory and in the current directory, but no luck.
The core would be in $PGDATA/base/yourdbOID/. If you don't see one,
it's likely because the postmaster is being started under "ulimit -c 0".
Restart it with "ulimit -c unlimited" and reproduce the problem and
you should get a core.
>> Also it might be worth your time to update to 7.4.2, just to see if this
>> is an already-fixed bug.
> I will look into that, but it would be non-trivial,
Shouldn't be; you won't have to dump and reload, just install new
software and restart postmaster. It's only cross-major-version updates
that are painful.
regards, tom lane