Service ignores notification delay and check attempts

[marc.wolff (marc)]

CMK version:
2.0.0p38
OS version:
Centos7
Error message:
Hi together,
some services that have notification delay and/or check attempts configured will send alerts on the first change of state.

Here is an example

Is that any kind of wrong configuration, or a bug? If i remove the service from cmk and discover this again, it will works. So that i think thats the configured rules are ok.

Any Ideas?

[robin.gierse]

This is on the remote site, and you spool notifications to the central site for delivery. How does it look on the central site? Also, you use periodic notifications, so maybe the notifications that is triggered is actually from an older event?

[marc.wolff (marc)]

Hi Robin,
Thanks for your reply.
The spool on the central and remote site is empty when I trigger a fake check result for testing.
I found this in the notification information:

My understanding is that it should be an OK state and not CRIT. Because the service was OK before I checked.
Is that the problem?

Thanks
Marc

[marc.wolff (marc)]

OK, the prev HARDSTATE is ok. I only change with my test only with my teast
But cmk should’nd notify in SOFT state.

[marc.wolff (marc)]

Helloo everyone,
for a quick fix, rebooting the host solves the problem. So it can’t be a rule or configuration issue.
We are planing the 2.1 and 2.2 update. Maybe it will solve the problem permanently
Thanks

[marc.wolff (marc)]

Hey again,
Unfortunately the problem still exists after the update to 2.1.
Now we are planning to migration to ALMA9 with new clear nodes.
I hope that if we just copy the confirguration, without any status and historical data, we will solve the problem.

Any other suggestions?

Thanks
Marc

[marc.wolff (marc)]

Hi everyone,
i have found a solution to this problem.
When I remove the service notification delay rule, the problem is solved.
When i create the rule again, it still works fine.

Is there a bug or conflict with this rule?
Can i access the failed rule access in the cmc.log with debug level?

Thanks Marc

[marc.wolff (marc)]

Hello everyone,

I have identified the root cause of the issue. While it might be functioning as designed, it appears confusing from my perspective.

Upon inspecting the debug logs from the notification helper, I observed that when the services transition from a “nok” state to “ok,” the notification helper still perceives the service as being in the “nok” state and logs the message:

“Postponing – notifications are disabled, but periodic notifications are enabled.”

The underlying reason for this behavior is our decision to disable the “notified events for services” rule, specifically the “Services recover to OK” rule.

Initially, I assumed this rule only affected notifications and not the core functionality of the CMC. However, it seems essential, especially when used in conjunction with periodic notifications.

Thank you. Mark