Service Performance graph didn't went down

[absarcompass]

Hi Team,
We face an issue where a UPS as host went down and services attached to it didn't went down. this happens because of global setting where services are not checked for when host is down.
Problem is when UPS as host went down the 'INPUT VOLTAGE' service kept plotting the graph at the value of 231 volts where as it should not be plotting graph at all as there are no new polling of services.
Is there a way where we can tweak this setting to have services attached to the host show no values in performance graph for the time host was down.

Host ping lost

SkyUPS-screenshot.png

Service performance graph

skyups-screen-serv.png

Cheers!

[tgriep]

I just have to verify that is this the global setting you have enabled in Nagios to stop service checks when the host is down?

host_down_disable_service_checks

If the above option is enabled, you may not be able to effect that service check as the global settings will ignore processing for that service when the host is down.

You could try and setup an Event Handler on the host to submit a passive check result for the service and see if it works but I am guessing that it may not work.
Take a look at this for more details on creating an Event Handler.
https://support.nagios.com/kb/article/n ... s-234.html

[absarcompass]

Yes, that is the global setting enabled in Nagios

host_down_disable_service_checks

My concern is if service checks are not working then why performance graph keeps plotting the last known value. if there are no new service checks then it should not plot any value, leave that period blank or something. Do we have any setting in performance graph which could achieve this?

Cheers!

[tgriep]

The Graph is truly not plotting when the service is not getting check.
What happens when the service starts up again, and it receives the performance data, it draws a line from the previous value to the current value so that is why you are seeing the straight line being plotted.
There is not a settings to change how this works except to let the service checks run when the host is down so it will plot the data.