Limiting notifications for a known service issue

[manymiles]

I'm monitoring my postfix queues using Nagios XI 2012R2.2.

Occasionally, I get back ups of mail for a reason that I am aware of but cannot control, for example, the MX destination for a foreign domain becomes degraded and some of my outgoing mail sits in the queue waiting for that destination to return but some gets delivered.

Therefore, my service state fluctuates between ok, warning and critical "frequently", but, not frequently enough to trigger flapping.

Since I am aware of the base issue as one out of my control, I'd like a way to NOT be notified of service problems for that service for a period of time.

I attempted to use the "Delay next notification", but, like it says on that page, it will not delay the next notification if the state changes. So, that didn't work. (As a secondary question, what IS the purpose of "delay next notification" if it does not actually delay notifications?)

I know I can use downtime to suppress notifications, but, I'm not really sure that it is appropriate?

Is there a better way to temporarily not be notified of a given service for a period of time other than downtime, or is that one way that downtime SHOULD be used?

Thanks for reading!

Miles

[slansing]

You can also acknowledge a service/host problem to suppress alerts. Delay Next Notification does work as intended. The issue you are running into is that when the state changes it resets the retry interval, and subsequently the notification withholding on that object.

[manymiles]

You can also acknowledge a service/host problem to suppress alerts.

Thanks for that. But, that too gets cleared when it changes state, so, I don't think I am any better off.

Any other ideas?

Miles

[slansing]

Well you can of course disable notifications on the object, or set up notification periods in which you will be notified. You can also set specific notifications to be sent, for instance if it commonly flaps between ok and warning, you could limit notifications to critical only. Downtime would work as well, though it does sound weird applying it in this situation it would serve it's purpose.

[lmiltchev]

Therefore, my service state fluctuates between ok, warning and critical "frequently", but, not frequently enough to trigger flapping.

You can change the low/high flap thresholds for a host/service to adjust the value according to your specific environment.

[manymiles]

Good suggestions. Thanks guys!

[slansing]

Sure, let us know how this works out, and if you are ready for us to close the thread.

[manymiles]

You can close it. I'm not sure when I'll get back to tweaking that situation.