Problem:

The functionality of total_number_of_zombie_processes has been broken as of Linux Kernel v.4.14 and the introduction of Process State: Idle.

This is due to a limitation in the library for net-snmp which makes it unable to distinguish between the process states: Idle and Zombie. The result of this limitation is that all Idle processes that are handled by the check_command "check_by_snmp_procs total_number_of_zombie_processes" reports all Idle processes as Zombie.

Solution:

If you still need to monitor Zombie processes, we suggest using NRPE. A guide on how to set up NRPE can be found here: Installation of NRPE agent on CentOS and RHEL.

One way to use NRPE to check for zombie processes would be to use the check_command: check_nrpe and specify:

$USER1$/check_nrpe -s -H $HOSTADDRESS$ -c check_proc_zombie

The check_proc_zombie in the above example is a NRPE-command.

The NRPE command check_proc_zombie needs to be added to the NRPE configuration on the host. This is specified in /etc/nrpe.d/op5_commands.cfg. Add the following line to the aforementioned file.

command[check_proc_zombie]=/usr/lib64/nagios/plugins/check_procs -w 5 -c 10 -s Z

Since we are defining the function on the client we need to pass it the arguments it requires, in this case -w (Warning) -c (Critical) and -s (State)

This will enable you to run the check_procs_zombie NRPE-command via the check_nrpe check_command in op5 Monitor and monitor Zombie processes.