On Sat, 10 Feb 2001 14:19:54 EST, Karl Signell <[log in to unmask]>  said:
> Every week or so, I get a frantic message from one of our listowners:
> LISTSERV is not responding!  I telnet in, issue a "go -bg" command and
> LISTSERV starts cranking out the backlog.  Can't find anything in the log
> to explain the outage.

This sort of sudden evaporation is typical of a resource limit
being exceeded.  On our machine, the 'lsv' process has been
known to grow to over 200M on occasion.  Things to watch out for:

1) System out of swap space - some operating systems will start
killing off processes if they are about to run out of swap.
If it's an "oldest biggest process first" algorithm, 'lsv' is
a likely candidate.

2) On AIX, the default data segment size is 128M, which kills
off the lsv process when it's on its way to 200M.  You may wish
to check what your system's default is.

3) I run 'go' out of /etc/iniitab, like this:

listserv:2:respawn:/home/listserv/go bg

The 'respawn' means if it dies off, we launch another one.
This does wonders for the number of times you get called at 2AM. ;)

                                Valdis Kletnieks
                                Operating Systems Analyst
                                Virginia Tech