Question about ping and downtimes in Icinga

#1

Hello everybody,

You might remember me from such ‘great’ posts such as ‘halp, my notifications are borked’ etc…

But this time I present a simpler scenario. I set my downtime for a host. I take the host down. It shows it’s in downtime, but the ping check still goes into a soft 1,2,3, then hard and sends notifications even if it’s in Downtime.

Any ideas why?

Icingaweb 2.6.1
Icinga 2.10.5-1

#2

Did you apply the downtime also to the services of the host?

1 Like
#3

Thanks log1c, I did not, no. I didn’t find too much info on what this really does but from what I understand there are already implicit dependencies, ie if a host is down and in downtime, then the services will not all be down as they have that dependency ‘link’ to the host. Thought that would be enough so would not have to use ‘All Services’. Hitting ‘All services’ means that the services are not checked (like the host?), if it’s unticked, there will not be notifications BUT the checks carry on being checked, or attempted to be checked. Not sure about that last part but that’s how I understood it.

As a side note, selecting ‘All Services’ also puts all of the services in the overview-downtime and if you want to cancel a host being down you have to go to every single service check in the downtimes and hit the red cross for that host. It’s an ok price to pay for having all this working though, but people ‘on the floor’ had better not make mistakes as it’s a click-fest after that.

TLDR: Is the common practice to use ‘All Services’ more often than not when creating a downtime? I’m not really bothered about individual services when doing this, the use will typically be to take down a host and all of its elements in monitoring so no notifications.

#4

Yes, there are implicit dependencies from services to their hosts. Also correct that, if a host is down, there will be no notifications for its services (regardless if the host has a downtime configured or not).

Downtimes only disable the notifications for host/services for that period of time. The check will run in their normal interval.

So when you set a downtime you have to think about the possibility that the host can be available while one of its services is not, due to some doing on the host. That’s what the “all services” tick is essentially for.

You can multi-select hosts and/or services and remove the downtime/comments/acknowledgements all in one go. Just use CTRL+Click or SHIFT+Click (like in the normal Windows/Linux desktop) to mark the objects.

1 Like
#5

Thanks a lot for the explanations and that tip at the end has blown my mind a little! Very nice, cheers log1c!

One thing I still don’t get though is that in my initial post I stated that i did not use ‘all services’ tick box, put in a downtime and then took host offline and none of the notifications were sent for services, but were for ping. So in effect, using ‘all services’ only stops the ping notification from being sent. Is ping (ping4 command) a special case? Just trying to get my head around… the logic :slight_smile:

#6

If I understand correctly the ping check is also a service (like cpu, memory, or whatever) of the host?

Can you share the object definitions of the host, ping check and the notification rule?
Normally there should be no notifications for any service belonging to a DOWN host.

I for myself have edited the icingaweb2 config.ini so that the “all services” tick is always set:
/etc/icingaweb2/modules/monitoring/config.ini

[settings]
hostdowntime_all_services = "1"
#7

That’s a great tip there log1c, thanks, have tested it out in our test environment and will use it in production.

Have managed to set a lot of downtimes for an operation last week and it works find with the ‘all services’ - had no issues at all. Was good to have the discussion and I learned some things, hopefully others will too from this thread. Thanks a lot!