Question & Answer
Question
Cause
If you are having problems with the Intelligent Management feature in the WebSphere web server Plug-in, gather this information and send it to IBM support.
Answer
2) What web server type are you using? What version? On what platform?
3) What are the symptoms of the "problem"? What is the URL that the user is trying to access? What error message does the user see in the browser? At what time did the problem happen?
4) Enable access to the /server-status page of the web server and change the LogFormat stanza in the httpd.conf file
- The Apache mod_status servlet must be enabled and access to the /server-status URL must be enabled in your http configuration file.
- Change the line
LogFormat "%h %l %u %t \"%r\" %>s %b %D %{WAS}e %X" common
to
LogFormat "%h %l %u %t \"%r\" %>s %b %D %{WAS}e %{RH}e %X" common
5) Set the Plug-in LogLevel to "Trace":
- In the WAS Admin Console go here:
Servers
Server Types
Web Servers
webserver1 (or whatever name of your web server)
Plug-in Properties
In the Plug-in logging section, set Log Level to Trace
Click OK and Save.
6) Set the Intelligent Management trace specification to ":DEBUG":
- In the WAS Admin Console go here:
Servers
Server Types
Web Servers
webserver1 (or whatever name of your web server)
Intelligent Management
Trace specification
In the first box at the top enter :DEBUG
Click OK and Save.
7) You need to Generate the plug-in config, and Propagate (copy) it to the web server.
Stop and restart the web server so the trace will show everything from the beginning.
8) Capture a trace of the REST service on the Deployment Manager
Set the trace file size and number of files per the instructions here.
Use the following trace specification:
*=info:com.ibm.ws.odc.*=all:com.ibm.ws.wsgroup.*=all:com.ibm.ws.xd.agent.odc.*=all
9) At the time of the problem, capture additional information:
Use the following command, on the Deployment Manager to capture the state of various Intelligent Management processes:
wsadmin.sh -lang jython -f [WAS_HOME]/bin/dumpIMPState.py --user=[WAS admin user] --password=[WAS admin user password] --auto > IMPState.txt
Save the web page output as an HTML file and submit to support.
10) After capturing the traces of the problem, please send the answers to the questions above and these files to IBM Support:
- web server configuration (ie. httpd.conf)
- web server logs (ie. error_log and access_log)
- plug-in log and configuration (ie. http_plugin.log and plugin-cfg.xml)
- Run the collector tool on dmgr process. When you run the collector tool, it creates a jar file with the information about that profile that it was run against, please send us the jar file that the collector tool created.
+ also send the IMPState.txt file and the server-status page created in step 9.
11) After the tracing has been completed, change Plug-in LevelLevel back to "Error", and remove the DEBUG trace specification from the Intelligent Management section. Also turn off the WAS traces by setting the trace specification back to *=info only.
NOTE: keep an eye on the size of the http_plugin.log file. It does NOT wrap like WAS log files. So, if the plug-in log is growing very large (over 2 GB) you will need to stop the web server, and delete or rename the file, then restart the web server. OR, you may want to consider using RotateLogs to automatically rotate the plug-in log to a new file based on time or size.
Related Information
Was this topic helpful?
Document Information
Modified date:
06 June 2024
UID
swg21681297