Every time we are restarting the IS or MWS , The connection is not getting created automatically with UM. We need to restart the UM every time. We get that the realm is not reachable.
One situation: 2 MWS in cluster. Restarting one MWS, that MWS is not able to connect to the UM. The other MWS can connect without any issue.
As soon as I restart the UM, the connectivity restores.
Same issue is happening for IS.
Same issue occurred for Triggers also after deployment of few packages. Only the triggers of those packages were not connected automatically after the deployment. Once the UM is restarted all the triggers got connected successfully.
*Error messages :
[6568]2023-08-21 13:21:54 [ISS.0134.0094E] (tid=943) Unable to start JMS connection alias JMSConnect. The server will attempt to reconnect every 20 seconds: com.wm.app.b2b.server.jms.JMSSubsystemException: [ISS.0134.9063] Error retrieving JNDI context: javax.naming.CommunicationException: JNDI setup failed on RNAME=nsps://abc:xyzy [Root exception is com.pcbsys.nirvana.client.nRealmUnreachableException: Realm is currently not reachable:Realm was still unreachable after max retry count - 0]
This totally sounds like an active/passive cluster or load balancer issue. Do you have any UM clusters in your environment? If yes what do you have configured in the VIP? Is there a F5 or HA cluster? Are they properly detecting dead nodes?
It sounds like cluster manager doesn’t register the dead node properly. It can also be caused because of a wrong VIP/DNS configuration in the endpoints. If reaching through a VIP only use that to access UM servers, never use hostnames in any configuration if you have a dedicated entry point.
You can test if this is because of the F5/HA cluster problem by changing your VIP name to host arrays like from nsp://umvip:9000
to nsp://umserver1:9000,nsp://umserver2:9000
You don’t have to test this through MWS or IS.
First use Enterprise Manager to test the VIP,
then kill the nodes while connected to see if it reconnects.
If enterprise manager connects without any problems then its probably a defect.
If enterprise manage doesn’t reconnect through VIP, then try connecting to the same active node directly with hostname from enterprise manager. If you can connect to the active node, then it means your cluster configuration is wrong. If you can’t connect to the new active node, then that installation in that node probably needs some patching/configuration or may be reinstallation.
