Memory issues on PI 7.1 - Java Engine

[Former Member]

Hi, we are having issues on our PI 7.1 server where interfaces progressively takes longer to complete every day. For example, on day 1 the sales will take a couple of seconds to finish processing, the next day 30 seconds, the next day 5 minutes and after 4-5 days it will take 2-3 hours to complete. The volumes are exactly the same. When the J2EE engine is restarted from SMICM, the sequence starts all over and the first messages only take a couple of seconds. As a workaround our basis guys have scheduled a restart of the J2EE engine every week, but I would like to find out what might be causing this. Our basis team setup alert monitoring and we are getting the following alerts from the system daily:

"Memory Info Usage Rate ACTIVE 96 % > 95 % Current value over threshold value"

Any ideas?

Thanks

Rudi

[Former Member]

Do you use custom java code? It looks like a memory leak.

Try setting the Garbage Collector to collect data more often.

[Former Member]

Hi Martin,

we do use custom Java code, but the same code was used in XI3 and PI7.0 and we did not have the same problems, the interfaces was upgraded to PI7.1. The only differences is that the code is now in a generic function library instead of the mapping own UDF and we are using the standard function RFC lookup function instead of using custom code to do the lookup. Is it maybe possible that the standard RFC lookup function is causing it?

Appreciate your feedback.

Thanks

Rudi