First, a bit of background:
One of our clients is experiencing spikes in CPU usage for WebSphere instances running one of our web applications (in other cases, other applications are great). They have a test environment and a live environment (both iSeries) that both experience a problem - with one application for each instance. We deployed this application locally in our own test environments, as well as for many other clients on iSeries without similar problems.
What is actually going on:
Every second or so, CPU usage for the WebSphere processor usage process occurs anywhere from 7%- 20%even if no requests were processed at that time. The client reported that peaks reach a level 30%. These spikes typically make up the 1.5%processor as a whole — other WebSphere instances typically use 0%— 0.1%in standby mode.
My research is still
So, I looked at the threads. One thread in a test environment used ~350CPU cycles per second. A similar thread in their living environment used ~1500CPU cycles per second (indicating that it has a larger processor). The call stack for these flows looks like
Type Program Statement Procedure
QLESPI QSYS 17 LE_Create_Thread2__FP12crtt >
QJVALIBJVM QSYS 7 startThread__FPv
J com/ibm/ws/util/Threa > run
J com/ibm/ws/util/Threa > run
J com/ibm/ws/util/Threa > getTask
J com/ibm/ws/util/Bound > poll
com/ibm/ws/util/BoundedBuffer. JVM Dump - , , :
Thread: 00002F82 Deferrable Alarm : 11
:
- - , ? (, !)
Deferrable Alarm? JVM Dump 4 . , , . WebSphere ( Windows) BoundedBuffer, , BoudedBuffer .- WebSphere , . , - ?
- telnet- , - , ? WebSphere ..? ?
- JVM Dump , , ?
, , .
30 (1)
, , ( , - ). Apache Axis. , ?!
30 (2)
, , - - , , - Apache Axis. , - . - .
, , "" - - , , , "" , .