[omd-users] log_initial_states impact in a dynamic monitoring environment

Michael Van Wesenbeeck michael+omdu at oxopia.com
Wed Jan 16 14:05:52 CET 2013


Hello list.

As my first post to the list, I'ld like to thank mathias and others for creating such an awesome product.

My actual question is about the log_initial_states config variable.
We are monitoring a cloud platform with lots of resources (and thus checks) being added and removed.
To add and remove the checks we use the .mk files in the ~/var/check_mk/autochecks/ directory, but we need to reload nagios to make it read the altered configurations.
We alter the .mk files because we rather not put a "cmk -IIu --checks=xxx HOST" in a script as this would remove checks for services that are unknown at the moment when we want to explicitly remove a particular service.
I've seen that every reload also logs the initial states.
I'm a bit worried about performance of SLA (statehist) functionality after a while when lot's of reloads are done.
I was wondering if there are other ways of adding services without logging the initial states. I was looking at the external commands that can be sent to the nagios command pipe for an "ADD_SERVICE" or similar, but I don't think it can be done. 
If it were possible to add a service in that way, I could have the nagios config rebuilt anyway so that the service is in the config for when a real reload or restart is needed.

Thanks in advance to shed some light on this.

Kind regards,
Michael Van Wesenbeeck


More information about the omd-users mailing list