Engine blocked without warning, needed restart (BW 5.13)

[Tristan Pierquet]

Hi all,

We recently encountered a problem with one of our Tibco process (BW 5.13, TRA 5.10).

Our process archive is running in Active-Passive mode, along with other process archives. We fine-tuned the JVM in order to get 2048Mo of Heap Space on both instances.

The process was comparing an XML file to its XSD reference file using an external jar (in order to optimize memory usage). Both XSD file and external jar have been extensively tested and were used since at least 2013 without problems. During this action, the process suddenly stopped advancing without logging anything. No core file has been generated, and there were no warning in the log file. The process archive also stopped consuming messages from the EMS. The instance was still shown as "RUNNING" in the Tibco Administrator, and the Hawk Agent did not raise any alert.

When we stopped the engine from the Tibco Administrator, the command was instantly executed and logged. Once the instance was restarted, messages were consumed and the initial process resumed from last checkpoint. Other instances and processes on the server were not affected.

Do you have any clue concerning this problem

Thanks a lot for any help you could give.

Regards,

[Emmanuel Marchiset 2]

You might be in a situation where your BW has lost connectivity with the JMS server.

I would suggest to check the following KB article :

https://support.tibco.com/s/article/Configure-BW-client-reconnections-with-EMS-Server