IBM WebSphere Application Server Performance Cookbook

WAS traditional runtime diagnostic trace script

  1. Download diagnostictrace.py to your deployment manager
  2. Review the runtime Maximum File Size and Maximum Number of Historical Files settings for the target JVMs to ensure sufficient trace will be captured.
  3. Execute diagnostictrace.py to set runtime diagnostic trace using one of the following options. Replace *=info with the desired trace string:
    1. Option 1: Set diagnostic trace on all alive and reachable JVMs excluding dmgr and nodeagents: 
      bin/wsadmin.sh -lang jython -f diagnostictrace.py --action set --trace "*=info" --serverType APPLICATION_SERVER | tee -a diag_wastrace.txt
    2. Option 2: Set diagnostic trace on all alive and reachable JVMs including dmgr and nodeagents: 
      bin/wsadmin.sh -lang jython -f diagnostictrace.py --action set --trace "*=info" | tee -a diag_wastrace.txt
  4. Reproduce the problem.
  5. Disable runtime tracing by using the same step above to reset the runtime trace to *=info.
  6. Upload the following:
    1. diag_wastrace.txt
    2. For each JVM:
      1. $WAS/profiles/$PROFILE/logs/$SERVER/System*.log
      2. $WAS/profiles/$PROFILE/logs/$SERVER/native*.log
      3. $WAS/profiles/$PROFILE/logs/$SERVER/verbosegc*.log
      4. $WAS/profiles/$PROFILE/logs/$SERVER/trace*.log
      5. $WAS/profiles/$PROFILE/logs/$SERVER/http*.log
      6. $WAS/profiles/$PROFILE/logs/ffdc/*
      7. $WAS/profiles/$PROFILE/logs/tpv/*
      8. $WAS/profiles/$PROFILE/javacore*txt
      9. $WAS/profiles/$PROFILE/heapdump*phd
      10. $WAS/profiles/$PROFILE/core*dmp