Mark Nickel <mnickel@dunsirn.com> writes:
> (gdb) info stack
> #0 0x080f6900 in SpinAcquire () at eval.c:41
> #1 0x0807875d in _bt_getroot () at eval.c:41
> #2 0x08079e0a in _bt_search () at eval.c:41
> #3 0x0807acab in _bt_first () at eval.c:41
> #4 0x0807926a in btgettuple () at eval.c:41
Hm. Don't think I believe this backtrace very much --- _bt_getroot
doesn't call SpinAcquire, at least not directly. Probably the next
step should be to rebuild with debug support.
However, before you do that ... it would seem that you have either
a stuck-lock problem or a corrupted btree index on one of the critical
system tables (pg_class, looks like). The locking issue could be
checked by restarting the postmaster to see whether the problem goes
away. If there's a corrupted index, you can use REINDEX DATABASE
to rebuild it --- but read the directions carefully! You need to do
it in a properly-started standalone backend, else you'll not get
far enough to do the reindex.
If neither of those things get you up and running, it's time to
roll out the debugger.
regards, tom lane