One of the process exceed soft / hard process limit on OS (CentOS 6.6; CentOS 6.7). Because high availability is required in this environment, I wrote a nagios-check script in SNMP-base. With this check nagios can verify process’s limit & the actual number of opened files by it. It is customizable, you can change limits & the monitored process.
The target was to create an informative check with minimal set of installed software / tool. For the check SNMPD configuration, script on monitored host & nagios configuration / adjustment is needed (some cases SELinux policy add also if this is the first check with SNMP-request). This check will display the monitored process name & actual opened file by it in number.
For more detail, please check the attached documentation, everything is defined on it (state-matrix of script; how to implement this check,…)