Ad Widget

Collapse

Cascading Zabbix triggers

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • flux
    Junior Member
    • Apr 2012
    • 3

    #1

    Cascading Zabbix triggers

    Hi!

    Please help me to decide what kind of logic I have to implement in the following scenario:
    1. I got some servers placed at 'location 1'
    2. I got some servers placed at 'location 2'
    3. Both locations represent a 'region'

    I need to compose triggers (I'm going to use this on my Map) to be notified if:
    1. location 1 (say data center 1) is dead, if all servers at location 1 are unreachable
    2. location 2 (say data center 2) is dead, if all servers at location 2 are unreachable
    3. region is dead, if both location 1 and location 2 are unreachable

    If you know about Amazon availability zones and regions, you'll probably know what I mean...
    It's sort of
    {smtp1.zabbix.com:net.tcp.service[smtp].last(0)}=0&{smtp2.zabbix.com:net.tcp.service[smtp].last(0)}=0
    from examples, but with dynamic host fileld I guess...

    What is the best way to do it?
    Last edited by flux; 16-04-2012, 21:37.
  • cirrhus9.com
    Member
    • Feb 2012
    • 58

    #2
    Have a look at http://www.zabbix.com/forum/showthre...9506#post99506

    Comment

    • Yello
      Senior Member
      • Apr 2011
      • 309

      #3
      Hi,
      I'll explain to you how I'd do this in my environment. I'm making assumptions that you're set up similarly enough to me for this to be workable for you:

      1. Monitor each host in Loc1 and Loc2 individually.
      2. Create an item for each host that determines whether it (the host or service) is alive or not.

      Note: I assume you've already done this.

      3. Create a virtual host - that doesn't tally with a physical server - maybe call it region "Region".
      4. Create two aggregate or calculated items, depending upon what you think will work best. The first one will measure service health in Loc1 and other Loc2.
      5. Create a trigger or triggers in line with the logic of the items created at step 4. The logic would be "if Loc1 has failed and Loc2 has failed then Trigger".

      Does this help you?


      Regards,
      David

      Comment

      • flux
        Junior Member
        • Apr 2012
        • 3

        #4
        Thank you guys for your advices!

        But how do I create an aggregated item for Loc1 or Loc2 that will be automatically updated if I add a new server in Loc1 for example?

        Assume I have Host1 and Host2 under Loc1, I monitor them by status key provided by Zabbix agent. How do I create an aggregated item in this case and will this item automatically see Host3 if I place it under Loc1?

        {<server>:<ItemKey1>.last(0)}#{<server>:<ItemKey2 > .last(0)} is just a trigger that keeps track of values of 2 servers <server> but can this rule be updated automatically somehow if I add <server3>?

        I actually need an item or a trigger where host/<server> value is not explicitly defined, so I can plug new hosts without editing this rule manually.
        Last edited by flux; 19-04-2012, 20:40.

        Comment

        • Yello
          Senior Member
          • Apr 2011
          • 309

          #5
          Originally posted by flux
          Thank you guys for your advices!

          But how do I create an aggregated item for Loc1 or Loc2 that will be automatically updated if I add a new server in Loc1 for example?
          By asking this question I can tell that you're not bothering to refer to the documentation. You should...

          An aggregated item looks at a hostgroup and checks each host in the hostgroup. I assume you have a sensible hostgroup configuration in place?


          Regards,
          David

          Comment

          • flux
            Junior Member
            • Apr 2012
            • 3

            #6
            I have figured it out with item aggregated checks (probably that's what you mean by aggregated items). Thanks for your help.

            Comment

            Working...