Not data collected on monitor with service dependency

[krobertson71]

Nagios XI 2.7
RHEL 6
NCPA 1.8.1 agents

I have just noticed something today.

We have backup processes on our Linux servers that run at various times throughout the week. In a lot of cases the CPU maxes out during this time and would start alerting.

I created a service dependency to correct this. HostA/CPU depends on HostA/backupprocess.exe

I have reverse logic with the backup process checks. If the process is seen running, that check goes critical and since the CPU check depends on it, alerting is stopped. But today I just realized that NO cpu data was being collected during this window. We have several now running for the last 8 hours and the cpu data is missing for that exact time window.

Here is the screenshot of the CPU graph:

cpudata_servicedependencies.png

Here is the service dependency rule for one of the hosts:

Code: Select all

define servicedependency {
       dependent_host_name                      servera.domain
       dependent_service_description            CPU Usage
       host_name                                servera.domain
       service_description                      Instances of save,Instances of savefs
       inherits_parent                          1
       execution_failure_criteria               w,c,
       notification_failure_criteria            w,c,
       dependency_period                        xi_timeperiod_24x7

[krobertson71]

Now that I have posted this and seen it laid out this jumped out at me.

Could this be the reason I am not getting data?

execution_failure_criteria w,c,

If that is it, assuming I would just uncheck that in the config.

If this is the case, please feel free to slap me for being stupid and move on

[BanditBBS]

Yes, if you just want notifications disabled while that one is critical then you don't need the execution_failure part. When the execution_failure part is included, when the checks are scheduled and the dependent service is critical, then the check just reschedules itself based on it's check time period and skips the current check, hence no perfdata.

[hsmith]

Yes, if you just want notifications disabled while that one is critical then you don't need the execution_failure part. When the execution_failure part is included, when the checks are scheduled and the dependent service is critical, then the check just reschedules itself based on it's check time period and skips the current check, hence no perfdata.

Thanks Bandit! krobertson71, did this work for you?

[krobertson71]

Yes it does. Like I said, once I read my post I like "Oh man I bet that's it".

Thanks Bandit for the help! Very much appreciate the answer and the detail explaining it

Hope you guys have a good time at NagiosCon this year!

[hsmith]

Yes it does. Like I said, once I read my post I like "Oh man I bet that's it".

Thanks Bandit for the help! Very much appreciate the answer and the detail explaining it

Hope you guys have a good time at NagiosCon this year!

Thank you, glad to hear this is working now. Going to mark this one resolved!