When adding a Java probe to a JVM startup command line of WebSphere, it is possible that there is a missing instrumentation in Diagnostics UI instrumentation charts. If there are problems during the launch of the probe’s process, it will not show active (green status) in the Health View menu from the probe. In case the probe service has started correctly locally and it appears in the Health View as active, it will be required to check the probe.log file located in the <probe_installation_folder>/log folder.
This guide show how to handle specific issues when instrumenting applications hosted in WebSphere.
Usually, when working with problematic Java probes and WebSphere the following error shows up:
The below steps are useful to fix these eventual instrumentation issues:
Check if the probe is working on top of the latest patches available.
Check if instrumentation defined in auto_detect.points can be viewed in the Diagnostics charts. If custom layers do not show:
Open server.err file from the log folder of the server for memory errors in the format:
log4j:ERROR Failed to flush writer, java.io.IOException: There is not enough space on the disk
If this is the case, instrumentation will have to be modified in order to consume less resources in disk.
Go to the "etc" folder on the Java probe, ie. <Diagnostics_Install_folder>\JavaAgent\DiagnosticsAgent\etc folder.
Open the file called "inst.properties" and increase the value defined by class.methods.max.size
Change the properties below in the etc\server.properties:
probe.pull.max.threads = 12
Search SEVERE error messages in the probe.log file, related to class or classes *.Bindings.*
This is related to access permissions to some private classes.
Edit etc/inst.properties file and search for line starting with the parameter classes.to.exclude and add the problematic classes.
Restart Application and Diagnostics Servers.