Nagios Support Forum

HMm,

I think I might have made a dashboard with a too complex query (something with topn of messgae field)

and it seems My NLS servers are both frozen atm. WHat do I best do to solve this? reboot? Kill Java process? Restart services?

Grtz

Willem

That query would be handled by elasticsearch, so one at a time restart elasticsearch, waiting a good 15 minutes between restarts.

Restarting elasticsearch service does seem to solve the issue, untill I go to that particular dashboard again lol. So, will I Have to delete that dashboard? Or can I edit it without starting / refreshing it?

Viewing that dashboard will kick it off, unfortunately. You could save it quickly to disk, edit the JSON to remove the offending panel, then reload it. Of course this relies on being able to save it quickly. A dev might be able to help more with doing this from the backend.

I managed to quickly delete the query causing java procs to spike. Isn't there some mechanism preventing cpu choking like this from happening?

Within Logserver and the associated ELK stack? I can't really say. I would like to imagine that Java itself would have some watchdog capabilities but even those can be faulty. Did you have something in mind when you asked this?

Lol Trevor, yeah some vague feature request thingy. Not really urgent, but it would be useful imho if for example a timeout could be configured that would stop the loading of the dashlets on the dashboard after x time (eg 120 seconds or so) . I'm not sure what is possible and what not. You can close this thread if you want Trevor.

WillemDH wrote:I'm not sure what is possible and what not.

I think we are in the same boat here. I started learning Logserver from the backend up, so I am not too familiar with what would be possible in the interface. My gut tells me this is still a backend issue, but AJAX is a hell of a thing and there might be a solution here. I'll put in that request.

Feature request 5110 has been filed regarding this issue.