Monit not detecting sleeping processes even though uptime rule triggers [closed]

I need to automatically end processes running for more than 24 hours on a CentOS 7 server. I’m using Monit to detect long-running processes and execute a custom script.

To test it, I reduced the threshold to 30 seconds and used a dummy process:

sleep 5000 & echo $! 

Monit configuration:

# /etc/monit.d/test_uptime.monitrc check process test_sleep matching "sleep" if uptime > 30 seconds then exec "/usr/local/bin/monit_kill_sleep.sh" if does not exist then alert 

Monit validates correctly:

sudo monit -t Control file syntax OK 

Monit also logs that the uptime rule is being triggered:

'test_sleep' uptime test failed for sleep -- current uptime is X seconds 'test_sleep' exec: '/usr/local/bin/monit_kill_sleep.sh' 

My helper script logs when it runs:

=== 2025-12-11T16:50:57 monit_kill invoked === no sleep processes found 

But the sleep process is definitely running:

pgrep -a sleep 235589 sleep 5000 

Monit calls the script, but inside the exec environment it is unable to see or kill the process. Even using a robust helper script with full paths (/usr/bin/pgrep, /bin/ps, /bin/kill) results in "no sleep processes found" when invoked by Monit. The script works when run manually.

Monit runs as root and has access to /proc. PATH is set inside the script and full paths are used, but Monit’s exec environment cannot see the process.

My question: Why does Monit’s exec environment fail to see or kill a simple sleep process, even though it is running and visible from the shell, and the uptime rule triggers?

What am I missing in Monit’s execution environment or process matching?