Websphere 4.0.5 / JDE SP23_k1 runaway thread I/Series

gcarmody

Member
We have just implemented SP23_k1 on an AS/400 running at OS/400 5.2.

We run AIO (Webserver/App/DB on one box) with Websphere at 4.0.5.

Since SP23_K1 we are experiencing severe intermittent problems with what appear to be runaway threads in Websphere. When the threads go runaway we also get near 100% disc arm utilisation on the AS/400.

The Web users grind to a halt and eventually we have to restart the Webserver subsystem.

The batch JDE and Fat Clients continue ok.

Has anyone encountered this before?

Any help appreciated.

Gerard
 
A colleague at another customer had this on the AS400 after applying SP23.

It was down to the garbage collection working overtime due to a large heap size setting.

This was not an issue prior to SP23.

They were recommended on advice from IBM this value has ben reduced to 512Mb which has stopped the huge disk arm utilisation issues - the CPU is slightly busier (presumably in doing more frequent GC) but not so that it causes any problem.
There heap size previously was 3Gb
Hope this helps.
 
Back
Top