Coherence*Extend Clients That Register Cache MapListeners Do Not Receive The Correct Number Of Events (Doc ID 1561014.1)

Last updated on OCTOBER 20, 2023

Applies to:

Symptoms

In the cluster an application has created multiple MapListener instances against a cache.  One hundred values are inserted into the cache with key values 0 to 99, then those values are updated and then deleted:

• Listener 1: Listener using an AlwaysFilter and the lite flag set to false. In other words, this listener should receive 100 insert, 100 update, and 100 delete events.
  
  
• Listener 2: Equivalent to listener 1, but with the lite set to true. This listener expects to see 100 insert, 100 update, and 100 delete events.
  
  
• Listener 3: This listener listens for one specific key value from the range being inserted. The lite flag set is to false.  This listener expects to see 1 insert, 1 update, and 1 delete event.
  
  
• Listener 4: Equivalent to listener 3 with lite flag set to true.  Similarly, this listener expects to see 1 insert, 1 update, and 1 delete event.
  
  
• Listener 5: This listener is similar to listener 1, but instead of an AlwaysFilter a LessFilter is used to receive events on keys less than 50.  The listener expects to receive 50 events for keys 0 to 49, which will translated to 50 insert, 50 update, and 50 delete events.
  
  
• Listener 6: Equivalent to listener 5, but with lite flag set to true rather than false. Again the same number of events is expected as for listener 5.

When the 100 objects are inserted, updated and deleted the listeners the number of events seen is tracked and reported:

The fix in 3.7.1.7 corrects the behavior of the listeners that are either registered to receive all events, or only events for objects that match an applied filter, but does not correct the issue for the listeners that have registered for changes to objects with a specific key value.

Changes

Cause

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.