Management of Multiple TIBCO MDM Instances with Caches

Multiple instances of TIBCO MDM use cache for exchanging job status and for distributed locks.

If this information is lost due to the abnormal shut down of TIBCO MDM instances or cache instances, the following events may occur:

  • Distributed locks may not be released.
  • Distributed locks may be released prematurely.
  • Jobs which are processing data batches may hang.

When you observe this type of abnormal behavior, shut down and then restart the whole cluster including all TIBCO MDM instances and cache instances.

To avoid such unacceptable situations:

  • Allow TIBCO MDM instances to shut down gracefully.
  • Set up replication for some caches.