4 Replies Latest reply: Sep 27, 2012 12:39 PM by Daryl E. RSS

    Notification - o/s script Environment Variables / "clear" signs

    809598
      OMS Version: 11.1.0.1.7
      Agent Versions: 11.1.0.1.7
      Repository DB: 11.2.0.2

      Hi there,

      Im having issues configuring custom o/s notifications.

      Currently when alerts are raised, they are propogated correctly to my ticketing system via o/s script. This is fine.

      The issue that im having relates to sending notification to clear these alerts. Let me explain,,,,,

      When an alert is raised and sent to the ticketing system, it is identified within that system by its cyle_guid. For example:

      - Alert is fired relating to "instance down" lets say, this alert has a cycle_guid od 'abc'
      -The issue is rectified and a "clear" message is sent from Grid Control again with cycle_guid 'abc'
      - At this point our system matches the cycle_guid coming from the clear notification and removes the entry from the ticketing system.

      However....

      Lets say the following situation occours

      -- A UDM threshold is breached for say "Tablespace X in DBTEST has only 10.1%", again the notification is sent (against this a cycle_guid say 'aaa')
      --The issue is not rectified
      -- Three hours later the metric is evaluated again "Tablespace X in DBTEST has only 9.6%" , notification is sent again (event though repeat notifications are turned off) - it sends a new cycle_guid - say 'bbb' and an additional record appears in the ticketing system.
      --The issue is then rectified at the point that last collection of this metric "Tablespace X in DBTEST has only 9.6%"
      -- Notification fires again sending a clear message again cycle_guid 'bbb'
      -- NO clear message is then ever sent against cycle_guid 'aaa' so there ends up being a record in the ticketing system that is never cleared.

      Another example

      -Alert fires for "1 invalid object schema x in db y" - this is assigned cycle_guid '111'
      - Another object goes invalid in the same db same schema - alert fires - "2 invalid object schema x in db y - assigned cycle_guid '222'
      - Another object goes invalid in the same db same schema - alert fires - "3 invalid object schema x in db y - assigned cycle_guid '333'
      - The 3 invalid objects are recompiles, the issue is rectified, - a clear message is sent against cycle_guid '333'
      - NO clear is sent against cycle_guid '111' or '222' - they remain in the ticketing system



      So my question....

      - What exactly does the cycle_guid represent and what dependency does its generation have on the other environment variables?

      - When an alert is sent and the metric has a variable key_value like the above example's (invalid objects/ tablespace free), what is the key or composite key for tying them togeather when "clear" is sent for the most recent?

      going back to the invalid objects example.
      when the alert for 2 invalid objects is sent, I need some way of identifying frmo the notification variables the previous notification of 1 invalid object, so that it can be cleared down,.

      Hope this makes sense,...
      Any ideas suggestions welcome

      Thanks

      Edited by: 806595 on 20-Jun-2012 14:25