What happens to an event that reaches the retry limit

  • 7011792
  • 15-Feb-2013
  • 15-Feb-2013

Environment

NetIQ Identity Manager Engine

Situation

Can someone please explain what happens to an event that reaches the retry limit? Do we just throw it away? A retry has a specific purpose in our guaranteed delivery of events. If we are going to throw away an event then how can we capture that this has occurred?

 

Resolution

Normally, we do not retry events.  When we do we only do it in a "custom" retry. Retry conditions are detected in policy and retries are managed in policy. The paradigm of guaranteed delivery does not fully apply to the DCS driver, because we cannot hold the event stream because a single event is failing and being retried. That's why we decided to back-off of having the driver ever issue a traditional "retry". The shim NEVER issues a retry but always an error.

What the policies do is:

- Detect well-known error conditions

- Retry the event causing that error condition by

- Removing the association of the object causing the offense

- "Migrating" the object causing the offense

- Appending and evaluating a retry counter to the @event-id

If the max retry count is reached, just leave the event alone, which means just report the error which originally triggered the retry.