Example - alert when number of sessions waiting to be indexed grows too large

About this task

When sessions are closed, they are moved from the Short-Term Canister to the Long-Term Canister, where they are queued for indexing. If the session indexer is unable to index the queued sessions fast enough, the queue can grow and grow. This unhealthy situation should be investigated.

In the following example, you can configure a canister alert to issue an alert when the number of queued sessions exceeds 1,000.

Note: For testing purposes, you may want to set this to a low value that you know is exceeded. Remember to reset the alert to appropriate threshold values after you tested it.

Procedure

  1. In the Discover Event Manager, click the Alerts tab.
  2. Select New Canister Alert.... The Add Alert window is displayed.
  3. For the Name of the Alert, enter the following:
    
    Alert - Sessions waiting for index > 1000
    
  4. Select a group for the alert, as needed. You might create a group such as the following:
    
    Canister - Spooling
    
  5. Select the Active check box.
  6. Click the Configuration step.
    1. For Alert Type, select Count.
    2. For the Event, click Select.... Select Un-indexed Sessions. Click OK.
    3. Do not configure a dimension value.
    4. Select the server. For this example, select Sum All Servers.
    5. For Direction, select Positive.
    6. For the Alert Threshold, set it to 1000.
      • You can specify the reset interval after the alert fired. If the interval is exceeded again before this time period expired, then another alert is not fired.
    7. If wanted, you can configure a warning threshold, which can be designed to fire when the alerting threshold is close to being exceeded. For this example, set the Warning Threshold to 750. For this example, leave the default value of 10 minutes.
      • You can specify the reset interval after the warning alert fired. If the warning interval is exceeded again before this time period expired, then another warning alert is not fired. For this example, leave the default value of 10 minutes.
  7. Click the Notification step.
    1. You can specify where you want to have the alert that is delivered and in the preferred format.
      • For testing purposes, you should send the alert to your email address.
  8. Click the Blackout step.
    1. If Canister maintenance is performed every night, you may want to schedule a blackout during the time of maintenance and for some time afterward, so that sessions that were queued during maintenance can be processed without triggering an alert.
  9. After you specified the alert, click Save Draft.
  10. The new alert appears in the Alerts tab in highlighted text.
  11. To commit the changes to the server, click Save Changes.
    • If you entered test values for the thresholds, remember to change the values to appropriate ones after testing is complete.
    • You may also want to revisit the Alert Notification properties.