load balancing and javanode still stand

[rainer_hbenthal]

Hi,

we have three instances of java running in our productional enviorment to get a load balancing.

From time to time one of this instances is stopping working on channels, whereas the others are still continously processing messages. Unfortunately there are still messages assigned to that "dead" node. AIX process control shows that this nod is still active, but it just stops processing messages for an unknown reason. Doing a soft restart if the J2EE Engine in SMICM makes everything working again, but it takes up to 10 minutes until everything is running again.

So, does somebody has a similar problem? How can i reactivate such a node without restarting the whole J2EE? And how can i detect that a specific java node is not processing messages any more? I guess there is no access from ABAP to check if one of the queues assigned to such a node is being filled up to configure an alert in CCMS?

Any help appreciated.

[Former Member · Accepted solution]

Hello,

Is it happening with one particular node ?

If so, then check the defaulttrace file of that particular node at the time the message fails.

You need to analyse the reason behind the node getting shut down.

Cheers.....,

Raghu

[rainer_hbenthal · Accepted solution]

Unfortunetly not, it happens to all three cluster nodes from time to time.

[Former Member · Accepted solution]

Hi,

That's weird.......

We could some hints on the possible reason of the issue only through logs such as defaulttrace and logs in the work directory.

So i request you to check the logs according to the time stamp of message failure or node failure.

Please post the same if there are any errors. I hope defaulttrace should catch the exceptions. Please check.

Cheers.....,

Raghu