Thanks for the information. Didn’t know that one yet.
Looking into that I think that this is related to the APT updates agent plugin provided by check_mk. Even with async it causes a disk IO spike.
Interestingly enough the script from LibreNMS that extends the SNMP run doesn’t create anything close to that high IO. So what is so different with the check_mk approach?
LibreNMS script for apt updates:
https://raw.githubusercontent.com/librenms/librenms-agent/master/snmp/osupdate
CheckMK info page for apt updates check plugin:
Every time the cmk apt plugin is run by the agent the disk spikes. Is this related to the cmk cache the info pages talsk about?
Why doesn’t the LibreNMS script cause anything close to this?