Monitor running time of a specific process

[freestyler]

I have been searching for a while on how to monitor a single process.

Basically, I wanna add an item that would be triggered when a specific process has been running for more than X time. (process duration time > 7h for eg.)

Is that something doable on Zabbix?
If not, is there any option?

[Calimero]

Have zabbix agent count the number of 'xyz' processes ? And then write a trigger like:

proc.num[my_proc].min(3600)>0

==> trigger is true if there was always at least 1 'my_proc' process over the last hour.

If you have processes 'my_proc' that are spawned every 2 minutes and last for 30 seconds, zabbix_agent will only check every X minutes (as defined in the GUI). So the trigger may think it was a single continuous run of my_proc while it was lots of shorter runs...

Another option if you're monitoring a batch / cron job is to write a small wrapper script and use zabbix_sender to send a state value to zabbix_server:

#!/bin/bash

$STATUS_STARTED=1
$STATUS_STOPPED=0

zabbix_sender ... -k myjob.status -o $STATUS_STARTED
/path/to/real_script.sh
zabbix_sender ... -k myjob.status -o $STATUS_STOPPED

And then write a trigger on myjob.status (probably some combination of nodata() and last()).

[freestyler]

Let's say I wanna trigger an alert when my process (my_proc) has been running for more than 4hours.

I'll write something like :
proc.num[my_proc].min(14400)>0

1/ how can I point to my_proc ? pid number ? how can this be automated ?

2/ Does this expression above work ? The way I see it is that it checks for the last 4 hours and see if my_proc is running (which is different since I wanna now that while I am checking that the process is running, and running for a while=4h )

[Calimero]

1/ how can I point to my_proc ? pid number ? how can this be automated ?

See documentation for proc.num.
First argument is process name (as show by 'ps' or 'top'). It can also filter on user, process state and command arguments.

2/ Does this expression above work ? The way I see it is that it checks for the last 4 hours and see if my_proc is running (which is different since I wanna now that while I am checking that the process is running, and running for a while=4h )

As I don't know exactly what check you want to achieve, my (fake) trigger was just an example.

If it's a scheduled task, you could also write a time based trigger ("if process is running and it's 10 o'clock" ...).