#230 procs shows incorrect count

[Anonymous]

On a RHEL 5.2 system, uxmon has the following directive:

localhost proc=ccd-traveler-admin.pl min=1 max=1 procs

Doing a "ps -efw | grep ccd-traveler-admin.pl" shows:

[cgi1 bigsister]$ ps -efw | grep ccd-traveler-admin.pl
ccd 25819 1 1 May13 ? 02:22:10 /usr/bin/perl -w /nas/cheetah-traveler/bin/ccd-traveler-admin.pl

However, BS complains that there are no processes running with that name:

Host Check Status Date Text
cgi1 procs red (1243018640) Fri May 22 14:57:20 2009
red ccd-traveler-admin.pl (no process)

I have observed this same behavior with some other files. It seems to be quite consistent. Most of the monitored processes show up without issue, but there are some (like this one) that error every time, on every host that is being monitored.

Another "problem" process name is admap.pl. That process monitor consistently shows no processes on every host that we try to monitor it on.

[Thomas Aeby]

As you can clearly see in your ps output, the process names in question are not ccd-traveler-admin.pl or admap.pl, thus they get a "no process" like this when you attempt to monitor them. Most probably, the process name is in both cases "perl".

In order to see what uxmon sees, take into account that BigSister will use the command "ps -e" on Linux systems in order to get a list of currently running processes. You can override this command using something like

localhost pscomm="ps -myoptions" procs

If you need to have process names extracted that are not quite written out by normal "ps" command, you can of course also use a script as "pscomm". Note however, that the last word on each line is expected to be a process name.

Best regards,
Tom

P.S.: I will move this request over to support requests, since this is not a bug.