• Introducing Dark Mode! Switch by clicking on the lightbulb icon next to Search or by clicking on Default style at the bottom left of the page!

Websphere problems!


Active Member
Hello List

We have EnterpriseOne 8.10, on Windows 2003, SQL 2005

We have recently rolled out to Web client and I have a Websphere Cluster which consists of 2 x virtual servers with 2 HTML Servers on each. We roll the client out to the user via a Windows 2008 R2 Remote App.
The Web servers each have 4GB Ram, and each JVM has a min heap size of 512 and a max heap size of 1024.

All was OK apart from the odd CPU issue where the CPU would spike for about 5 - 10 minutes on one of the Java.exe processes. I put this down to a large find somewhere on a poorly written app. As the web servers are virtual with only 1 Processor assigned our I.T Guy agreed to add an additional CPU to see if this would help with the CPU spiking issues (until we found and fixed all the problem apps).

Since we have added an additional CPU the HTML Servers CPU still spike, only now the JVM restarts itself and anyone logged into that JVM has to logout and back in again. In the SystemErr.log we see a lot of java.lang.OutOfMemoryError messages.

Tonight I plan to remove the extra processor that was added, however does anyone have any ideas as to why this would happen? Does Websphere need any setting changes to make it aware of the extra processor?

Thanks in advance



Legendary Poster
I believe each java.exe process will only utilize 1 CPU at a time, so if you had 2 instances on the VM, then each java.exe would theoretically use different vCPUs.

This, of course, does nothing for the out of memory messages. You might want to think about increasing the heap size to 1536.


Active Member
Yes, I have put the JVM's on 1 server up to 1536 and on the other I have removed the extra CPU. So I guess I will see tomorrow what happens. By the way do you know what the limit is for the Java Heap size?

B Scott

I believe the CPU spike is related to Java recording the heap dump just before the restart. Since the spike is caused by a single process, it would only use one CPU. Having two CPUs would allow the other JVMs to continue running during the minutes that process might take.
I got the same problem and I put 1024 min and max in the heap size , and now not see that message in jas.log , but one of my java.exe take a lot of memory and suddenly the system crash , yesterday I have to reset the services twice , I still looking for the solution of that

Windows 2008 32 bits
SQL Server 2005
EnterpriseOne 8.11SP1 Tool Release