4

I have created several hostgroups in nagios and each hostgroup consists of several hosts. Hosts carry applications which are monitored by service checks, always at least 7 or more per application. Thus my problem is: when lets say application on host A in hostgroup "Testing" is going down I suddenly receive about 7 notifications describing that every of 7 controls of application on host A in hostgroup "Testing" is in critical state.

What I would like nagios to configure is to send one notification that service check of, for instance, message count is in critical state on host A and then maximum of one more notification that hostgroup "Testing" is down. This way I do not get notified 7times and I know that I have to fix problems on specific host in specific hostgroup. This way it is more clear what problem to solve.

To add another example: When application on host A goes down and I receive lets say 10 notifications, few seconds later http_checks will start to notify me as well because apache does not recevie data from application which is down. So I end up solving one problem and receive about 20 or more notification. What I would appreciate is maximum of 4 notifications. One from one service check on host A and one per hostgroup in which host A is and then the same per hostgroup where is apache. Or if they are in the same hostgroup there would be just 2 notifications at all.

If similar problem occures in another hostgroup at the same time I would again know that there are two hostgroups with problem to fix. Current situation however is that I receive about 50 notifications and get confused where to start and what the real problem is.

Is anyone of you facing similar problem? I was looking quite a long time to any similar topic to solve the problem. I tryed to use dependencies but did not find way to configure nagios to the situation I described above. Parent - child relationship can be used only per hosts. Escalation does not solve this problem at all.

Maybe I just missed some information in documentation regarding this configuration. Would appreciate any advice.

Lucius
  • 41
  • 1
  • 3

1 Answers1

2

Nagios can do this with Service Dependency checks. See: http://nagios.sourceforge.net/docs/nagioscore/3/en/dependencies.html

But it's a real pain to set up and keep managed. I found the simplest solution was to use the fact that NRPE commands (defined on the target host), can execute a nearly unlimited number of actual checks. All via a single Nagios service check. I 'bundle' all the checks for a single application (process up/down, various log scraps, log ages, etc.) so that each individual applications has only a single check. The check results tell you which check has failed.

Jim Black
  • 1,422
  • 1
  • 13
  • 26
  • I actually really like this idea. This also would help prevent getting spammed by Nagios alerts. – Lance Badger Nov 03 '14 at 17:46
  • This kind of robs you of the flexibility and the information richness of using different services for different checks. Nagios is very powerful and can handle a lot of customization and configuration, and shame to avoid it. – jwg Mar 01 '16 at 08:15
  • 1
    I agree that Nagios is a very capable monitoring system. Its biggest fault isn't in what it can do, but how it must be done. Many modern monitoring systems have full GUI interfaces which allow operations teams to manage checks with mouse clicks and drag-n-drops. Setting up complex inherited/dependency checking in Nagios .cfg files is a nightmare for most people. – Jim Black Mar 03 '16 at 21:08