Top ten best practices for WebSphere administrators

Body

The following list consists of the top ten best practices compiled from numerous common problems that I have seen throughout my technical support years. Most of these issues could easily be prevented by taking these simple precaution steps.  So, if you're a WebSphere administrator, this is your must-read!

1. If you're on Linux or UNIX environment, make sure the ulimit value for no-files/open files or file descriptors is 10,000 or higher.  This is the recommended value for migration tasks, application deployments, synchronization, and so on.  There are so many problems associated with this value being low. See the following technote on how to set your ulimit value: http://www.ibm.com/support/docview.wss?uid=swg21469413
    
2. Use the latest install tools such as Installation Manager for WebSphere Application Server Version 8 or later, and Update Installer for WebSphere Application Server Version 7 to perform upgrade and /or install. Applying the latest version can avoid known issues that are associated with the tool being old and out of date. Check out the WebSphere Support site for the latest updates and downloads: http://www.ibm.com/support/entry/portal/product/websphere/websphere_application_server
    
3. Remember to clean up the cache (workspace folders) once in a while. Sometimes, clearing out old cache helps relieve the workspace disk space and mitigate some common problems such as start/stop server problems, synchronization, and so on. See the following technote for instructions on clearing out caches: http://www.ibm.com/support/docview.wss?uid=swg21607887
    
4. Spend some time to make a system backup of the WebSphere environment. The backup can become very handy when you encounter a serious problem and cannot recover. It will save tons of time from rebuilding the environment especially if you're in a production environment.
    
5. Always take a backup before editing a file, especially if you're making changes to a configuration file under WAS_home/config structure. Most of  these files are unique to the environment and WebSphere Support may not have a working sample to replace in case it got corrupted and causes problems.
    
6. Don't forget to make a backup of Installation Manager data folders such as IMShared, and IMData to prevent corruption and out of synchronization problem.  Again, having these folders backed up can avoid time spent on rebuilding the environment and help with recovery.
    
7. Check your heap size and ulimit for no-files/open files/file descriptors when deploying an application, especially if it's a big EAR file size. Additionally, make sure the user has the correct permission with read/write to WAS_home directories.  As a rule of thumb, the default heap size of 256 MB can be too low and may need to increase to 1024MB for a 64-bit environment. See the following technote on how to increase the heap size: http://www.ibm.com/support/docview.wss?rs=180&uid=swg21417365
    
8. Grant Permission, permission, permission.  By default, WebSphere_home and its subdirectories should have file permission of 755 (read/write).  Therefore, it's relevant to be consistent with using the same user ID (the same user ID that was used at the WebSphere installation) when managing WebSphere. Optionally, you may apply non-root user, but it has to be kept consistent.  This is also very important if you're using Installation Manager. See the Knowledge Center for more details on non-root installations for different operating systems:
   http://www.ibm.com/support/knowledgecenter/SS7JFU_8.0.0/com.ibm.websphere.express.doc/info/exp/ae/tpro_manage_nonroot.html
    
9. Most of the times the problems are not be due to WebSphere but because of the network!  Things to double check are:
   • Make sure that the important ports from WebSphere profiles (serverindex.xml) are open on the firewall.
      
   • Ensure that the host name/host names are correctly defined in the serverindex.xml file and and also in the etc/host files. Verify your connection. Are they pingable? Does a telnet test work?
      
   • Use unique port numbers to avoid port conflict. Here are the important ports from WebSphere perspective:
     http://www.ibm.com/support/knowledgecenter/SS7JFU_8.0.0/com.ibm.websphere.migration.express.iseries.doc/info/iseriesexp/ae/rmig_portnumber.html
      
10. Last but not least, if you're unsure what wsadmin command to use or how it use it, we have a wsadmin command assistance tool available on the Admin Console that you can turn on and run through the task on the console.  The command assistance window then would display the wsadmin command in Jython of the last task that was executed. Here are the instructions on how to turn on command assistance:
    http://www.ibm.com/support/knowledgecenter/SS7JFU_8.0.0/com.ibm.websphere.express.doc/info/exp/ae/ucon_rpref.html