JAS instance hangs

wadie

Well Known Member
Hi all,
Since october we are running 5 JAS instances (for 5 environments) on 1 JAS server. Only 1 instance, our PD instance, is used pretty often (15-20 people on it all the time)

In the beginning we didn't have problems, but lately the PD instance hangs frequently. It has happened 2 times today and over 5 times this month.

When the instance hangs, it looks like the browser keeps loading the page. Same thing happens when trying to connect to the instance using SAW. I don't think it returns an error, even if you keep waiting.
We can't see anything in the logs (not JAS logs, not WebSphere, not IBM HTTP) that could lead us to a reason and/or a solution...

The only thing we can do to solve the problem is restart the services each time it happens... We also have to kill some java.exe processes manually when the services are stopped.

Did anyone experience a similar problem? Did you find a solution?
All our instances have the same setup, so is the load causing this? (all other instances are barely used)

JAS server on IntelNT, SP23J_B1
 
Hi,
I don't know if the problem is the same, But with WebSphere 4.0.5 you must install a fix downloadable from IBM site regarding an error "Internal error 500" received by users during navigation on HTML interface (or install WebSphere 4.0.6 where this problem is solved).
Good luck
Gigi
 
Hi gigi,
we installed a PTF for WS and I think I remember that it was the PTF you are talking about (PTF 5 says the zip file)
 
I don't know if this is applicable however we had a client running Websphere 5 having a similar issue. The problem in that case was a heap size being too small. We increased the heap size from 256MB (the default size) to 768 MB and the problem went away. BTW... Peoplesoft recommends not running so many instances on a single server. (or at least the guy at GSS we have worked with in the past recommends it.)

-Paul
 
Hi Paul,

Heap size still is the default 256MB, so maybe I should increase it. This will give problems if I am going to increase it for all instances though, so maybe we should start looking for an extra server...
I'll start by increasing the heap size for the instance that has problems and see how it goes
 
Back
Top