So, it seems that I found the answer. :) As you have written: "if the host is down the service checks are suppressed". This would be the correct working situation, if is everything OK with the configuration. But, there is an excellent topic about this case:
http://sites.box293.com/nagios/guides/configurations-and-definitions/host-and-service-check-intervals It is unnecessary to define any dependency configuration, just normalize the checking intervals, because in ordinary case, if the Icinga knows, that a host is down, either of associated service checks will perform. The problem is when service check reaches earlier the hard state than host check and implies a notification, before Icinga can be aware of host's unreachability (down state). I have mentioned, that the whole configurations is not my creature, I have just inherited it from my ex-colleague. It seems, the check intervals were incorrect and I have to think about it and fix them. :) Thank you for previous answers! Kind regards: Gergo 2016-07-14 15:42 GMT+02:00 Szabó Gergő <[email protected]>: > Hi Brian! > > 2016-07-14 15:14 GMT+02:00 Brian O'Neill <[email protected]>: > >> I'm not talking about the host dependency, but the host check itself. >> > > I know. :) > > >> Did you receive a host down notification? Are you configured to receive >> one if the host is down? >> > > Yes, previously I have no problem with it. > > >> Did the logs show that the host is down? >> > > Yes, but it seems, that it reached only the second SOFT check: > [2016-07-14 10:36:55] HOST ALERT: myHost;DOWN;SOFT;2;(Host Check Timed Out) > > A host down notification should came due to device reload, but it can be > possible, that check_interval is too high to perceive if a device is > unreachable (is in host down state) - check_interval = 5min, retry_interval > = 2min. > > >> Did the pingv4 check reach a HARD state? >> > > Yes. > [2016-07-14 10:34:15] SERVICE ALERT: myHost;PING_IPv4;CRITICAL;HARD;1;PING > CRITICAL - Packet loss = 100% > > >> >> How it is working on my systems is that if the host is DOWN (or >> UNREACHABLE) and HARD, all the services stop getting checked, and will not >> reach HARD state. Thus no notifications for the services will go out, so >> long as the check interval of the host isn't longer than the check/recheck >> interval of the services. >> > I don't use a ping service for service dependencies in that way, since if >> the host is down the service checks are suppressed. I do have a ping >> service check for statistical purposes (my host checks only do a couple >> pings, while the service check does more to better measure loss and RTA). > > > Till I didn't configure dependencies, when a device went down, all of > related services notified too, this was the main problem. :( Unfortunately, > I inherited the whole Icinga configuration from my ex-colleague, so now I'm > on the way to fix some problems and configure new Icinga2 from the > beginning. :) > > > > Gergo > >
_______________________________________________ icinga-users mailing list [email protected] https://lists.icinga.org/mailman/listinfo/icinga-users
