Hi Tom:
No way to get the core dump, this is what I did:
1) Add "ulimit" to /etc/rc.d/init.d/postgresql just before pg_ctl start
in the "start" section of the script:
...
ulimit -c unlimited
su -l postgres -s /bin/sh -c "/usr/local/pgsql/bin/pg_ctl -D $PGDATA
-p /usr/local/pgsql/bin/postmaster -l $DIRLOG/logfile start > /dev/null
2>&1" < /dev/null
...
I tried "man ulimit" but it didn't work, "man bash" returns the command
help.
2) Restart postmaster
/etc/rc.d/init.d/postgresql restart
3) Run the query that produces the crash.
4) Search core file:
-bash-2.05b$ find /usr/local/pgsql/ -name '*core*' -print
-bash-2.05b$
Nothing comes up.
Thanks for your help.
Ruben.
Tom Lane wrote:
ruben <ruben20@xxxxxxxxxxxxx> writes:
As I recall, released versions of PostgreSQL usually dump core under
$PGDATA/base/<database oid>. However, it's also possible that your
coredumpsize resource limit prevents core dumps; you could fix that
by putting a command like "ulimit -c unlimited" in your PostgreSQL
startup script and then stopping and restarting PostgreSQL.
I cannot find a core file, ulimit is set to unlimit. I guess I'm doing
something wrong:
-bash-2.05b$ ulimit
unlimited
-bash-2.05b$ find /usr/local/pgsql -name '*core*' -print
(1) The fact that it's unlimited in your user environment doesn't prove
that it's unlimited in the environment the postmaster is started in.
(2) I forget which constraint ulimit-with-no-argument prints, but it's
not core file size. (Try "ulimit -a")
Please actually follow the advice given to you above.
regards, tom lane
---------------------------(end of broadcast)---------------------------
TIP 2: Don't 'kill -9' the postmaster
