In the October 2004 thread "postmaster stopps" Tom Lane et al.
suggested better logging to capture error messages when postmaster
silently stops after starting. I was getting silent stops whenever
booting the host, so I tried to follow that advice. I found that
redirecting the standard and error output to a named file instead of
/dev/null didn't merely capture the error messages, but stopped the
error from occurring in the first place. So, I appreciate that advice
greatly.
This is under the current version of Red Hat Enterprise Linux AS
running PostgreSQL 7.4.5, starting it at boot time with a custom
version of Ryan Kirkpatrick's /etc/rc.d/init.d/pgsql file.
---------------------------(end of broadcast)---------------------------
TIP 4: Don't 'kill -9' the postmaster
