Murthy Kambhampaty <Murthy.Kambhampaty@goeci.com> writes:
> Our postgresql server crashed this morning unexpectedly. Attached is our
> server log from before the crash. What's perplexing is the sequence
> "2002-08-01 10:17:30 DEBUG: server process (pid 13715) was terminated by
> signal 11
Core file? Stack backtrace?
> 2002-08-01 10:17:30 DEBUG: database system was interrupted at 2002-08-01
> 05:30:41 EDT"
> But there is nor report of any problems at 05:30:41 EDT.
I believe that's effectively the time of the last checkpoint. If the
system was idle between 5:30 and 10:15 or so then there's nothing very
surprising there. Tell us about the core dump.
> I should note that we deployed this server last week with
> checkpoint_segments=3 and fsync off. The server crashed repeatedly on file
> imports when the checkpoing interval dropped below 5 secs.
Define "crash". Again, stack backtraces would be useful.
FWIW, a lot of the reports of instability we've seen recently have
traced to hardware problems. How confident are you of this hardware?
regards, tom lane