Icinga2 on Windows has Problems with the execution

Hi There,

I recently did some performance troubleshooting on a Windows Server where the Icinga2 Agents and its Plugins seems to crash the Server.

I found out that it does not matter which plugin or even powershell scripts where executed, all the execution was arround 30 sec, instead of normally 0,1 - 1 sec.

Then I realized that a antivirus software was in place, I disabled it for testing and it worked immediately fine.

Than I checked if there where other servers that hat longer execution time and yes, there are. With different antivirus scanning software in place. I was wondering because I never found this kind of problems yet.

What do you think about:

  • Will it maybe make a difference if the service is running as system or as network service?
  • Does the Icinga2 Agent dir always has to be in the exclusion list? I never did with nsclient either…
  • Its not always. Its happens just sometimes and I cant really tell when it will happen.

Does anybody else ran into this kind of problems yet?

I tested with many different Windows Versions of Icinga2 (2.7.0 - 2.8.4)

Thank you for any hints on this!

Will it maybe make a difference if the service is running as system or as network service?

That very much depends on the plugin, your usage of them and the environment. If everything works as network service you should leave it that way.

Does the Icinga2 Agent dir always has to be in the exclusion list? I never did with nsclient either…

Please elaborate

About AntiVirus: Now and then there are people reporting problems with Icinga 2 and AntiVirus. Icinga 2 and especially the check plugins use a lot of system utilities and put their fingers where a non-monitoring software would not never, so it’s not surprising for an AntiVirus to get in the way. Can you “trust” Icinga 2 in your antivirus solution?

1 Like

For one customer I created the execlusions but I still see a longer execution time. Now I noticed that many systems schedule all the service checks ( 30 - 60 per server) at once.
I thought that on icinga2 service start the scheduling of each check is load balanced so that not every check is executed at the same time?
I have seen this on different Windows Icinga2 verions (2.7.0 - 2.8.4).
Am I wrong at this point?
All the checks are executed on each server locally.

thanks.

Padding should be done and increase over time. With 2.9.0 the checks will also be better balanced after a restart.

Is there something I can do or check until the 2.9.0 version is released? I know its currelty planned for wednesday but it will take me some time to update all the icinga2 windows agents…

Thanks.

Wednesdays is not fixed and likely to be pushed again, there’s a new bug opened discovered 3 hours ago. Not much you can do here, and we all know that everyone is waiting for 2.9 - still, one cannot release a broken version.

1 Like