On Wednesday February 16 2005 8:54, Tom Lane wrote:
> "Ed L." <pgsql@bluepolka.net> writes:
> > Postmaster is totally
> > unresponsive to SIGTERM, SIGINT, and SIGQUIT (sigkill is
> > the only thing that shuts it down).
>
> gdb backtrace definitely sounds like a good idea. Also it
> might be interesting to see the results of strace of the
> postmaster while hitting it with signals. Find out if it's
> doing anything at all in response ...
Workin' on gdb/strace build, but having trouble with both. Gdb
just doesn't build from source, not sure why yet, and no depots
found at porting center. It doesn't appear that strace is
supported for 11.23 itanium based on strace README/PORTING files
and 'man 2 ptrace'. HPUX users, is that your experience?
Question: Am I doing all I can to avoid corruption with the
following procedure to shutdown a 7.4.6 cluster with a hung
postmaster? Suggestions?
1. SIGTERM non-postmaster processes
2. SIGINT non-postmasters
3. SIGQUIT non-postmaster
4. SIGTERM postmaster
5. SIGINT postmaster
6. SIGQUIT postmaster
7. SIGKILL postmaster
TIA.
Ed