Websphere not restarting after 9.24 agent JVM parameter added

we have an application running in websphere applciation server version 7 with JRE 1.6 in solaris 10, we are able to install 9.20 agent, but we need to install 9.24 (either upgrade or new installation) but either way, we are trying the following steps and trying to invoke the agent but application server is not restarting.

following steps we did


1) we extracted .tgz file 

2) we installed running setup.sh

3) installtion and validation was successful (we didnt got any error message)

4) After that we run jre instrumenter with websphere JRE path and w got the path that need to be copied in JVM startup arguments

5) Then we copied the JRE instrumenter path to the JVM arguments and when we try to restart the server, its not starting,


Since the agent is not even invoked (we are not getting any logs).

I am attaching all logs and details that i have send to the support people in HP.I hope the experts here can solve it much faster





Parents Reply Children
  • Is there folder created such as "MercuryDiagnostics\Server" after installing the Java Agent?


    Is JAVE_HOME variable is set?


    Sometimes if the instrument is not correct then Application won't start. You check for exact problem in the Application logs.




  • Kmadan, Thanks for reply


    There is no logs related to that in WAS logs. after removing the diagnostics parameter and if we are restarting the JVM, then it is showing logs.else nothing is showing up.

  • there is no such folder inside mercurydiagnostics.Also we are not using any startupscripts, we are doing instrumentation manual mode and copying the JVM arguments in the JVM administration console.As i said, 9.20 is working fine after instrumentation.But we need to install 9.24.


    btw while doing the first step of extraction of .tgz file, we are seeing some @longlink lines inside the log. It looks like below.will it cause any problem? 


    gzip -dc HPDiagTVJavaAgt_9.24.113.139.tgz | tar -xvf -
    x JavaAgent, 0 bytes, 0 tape blocks
    x JavaAgent/DiagnosticsAgent, 0 bytes, 0 tape blocks
    x JavaAgent/DiagnosticsAgent/classes, 0 bytes, 0 tape blocks
    x JavaAgent/DiagnosticsAgent/classes/boot, 0 bytes, 0 tape blocks
    x JavaAgent/DiagnosticsAgent/classes/boot/mercury, 0 bytes, 0 tape blocks
    x JavaAgent/DiagnosticsAgent/classes/boot/mercury/picocontainer, 0 bytes, 0 tape blocks
    x JavaAgent/DiagnosticsAgent/classes/boot/mercury/picocontainer/Startable.class, 158 bytes, 1 tape blocks
    x JavaAgent/DiagnosticsAgent/classes/boot/mercury/picocontainer/PicoVisitor.class, 395 bytes, 1 tape blocks
    x JavaAgent/DiagnosticsAgent/classes/boot/mercury/picocontainer/Parameter.class, 704 bytes, 2 tape blocks
    x JavaAgent/DiagnosticsAgent/classes/boot/mercury/picocontainer/alternatives, 0 bytes, 0 tape blocks
    tar: ././@LongLink: typeflag 'L' not recognized, converting to regular file
    x ././@LongLink, 118 bytes, 1 tape blocks
    x JavaAgent/DiagnosticsAgent/classes/boot/mercury/picocontainer/alternatives/ImplementationHidingCompo, 3053 bytes, 6 tape blocks
    tar: ././@LongLink: typeflag 'L' not recognized, converting to regular file
    x ././@LongLink, 115 bytes, 1 tape blocks



  • Do you mean the log is empty? What about enabling debug logging? 


    It should create some log to tell what caused the failure to start WAS. I haven't see an instance where WAS doesn't generate any logs related to failure to start the server.

  • @KMadan We are checking the startserver.log , in that we are not recieving anything, Log is enabled, because if we remove the parameter and restart the server , at that time we are recieving all logs that corresponds to server startup. Is there any other specific log file we need to check? 


    Btw can you look into that extraction log lines, will it cause any problem?