Memory leaks caused by a websphere java process

We are on AIX 6.1 with Oracle 11g and using WebSphere 7.0_FP31.
Lawson versions 901_ESP12 and 901_MSP9.

We have AIX6.1 on an LPAR.

Websphere is clustered (2 apps servers....server1 & server2).....and since we went to WAS_7_FP31 we have had massive Memory Leaks that cause us to not only restart the Lawson Services, but to ultimately reboot the server to get it back to normal, only to find that it does not take long to get back to same problem once paging is being used quite heavily.

Anyone else have this issue?

Note: The clustered WAS instace does not appear to be built correcty and I know it may have to be scrapped and rebuilt using a total of 3 app servers in teh cluster instead of only the two. Personally I have never clustered a WAS instance with less than 3 before and not sure who did it this way or why. I have a case opeed with GSC due to not being able to start the ENterprise service for ios-cluster after creating it per a document found on the KB.

Find more posts tagged with

Infor Lawson Technology Group - Discussion

Comments

ludovic-masuyer

Note: Paging space being consumed is the trigger.

Vid dEPM-652 - Custom Report.mp4

serafino

I have never run WAS in a clustered configuration, but I am very famliar with its memory leaks. On Solaris, I have seen the WAS Java image grow to as much as 10G even though the heap size is set to max out at 2G. This will happen in a single server configuration if WAS is allowed to run for a week. Dark Majick applied to the JVM shows the actual heap is only 2G, but the runtime continues to allocate memory until it runs out. Interestingly enough, this never happens with the Lawson services that run "real" Java.

Copyright © 2025 Infor. All rights reserved.