Slow Application Server Response or Hang in Network Deployment Environment
 Technote (troubleshooting)
 
Problem(Abstract)
WebSphere® Application Server takes a long time to start 23 to 26 minutes. No errors are being logged in the systemout..log file. The server may appear to be hung or not responding. The cause identified in 2 and 3 below are relevant prior to Application Server V6.0.2.1.
 
Cause
1. Excessive J2CA0075W errors being logged in the systemout.logs for the server.

2. .Dormant activity.log.lck file in the /LOGS subdirectory at the server level

3. Activity.log file set to different sizes in the node agent and servers sharing the same activity.log file.

 
Resolving the problem
1. If the systemout.log file for the server(s) show excessive J2CA0075W errors, an immediate workaround for the problem can be found at the following link. The cause of the errors should be investigated by IBM support. http://www.ibm.com/support/docview.wss?&context=SSEQTP&q1=1231761&uid=swg21231761&loc=en_US&cs=utf-8&lang=en
2. The activity.log.lck, the activity log is a shared file. WebSphere uses it to implement a cross-process lock using a lock file (activity.log.lck). Normally, the JVM process will create this lck file when a message is written to the activitiy.log. The lck file is removed when the write is complete., When a process cannot access the lck file, it will wait for a specified time, then retry up to 5 minutes. After 5 minutes, it then attempts to delete the file and write the entry. If a server crashes before the process completes writing the message, it leaves a dormant activity.log.lck file in the /LOGS subdirectory for that server. There are a couple of suggested workarounds for this situation.

Check the /LOGS file for any server that previously crashed to see if an activity.log.lck file exists. If so, delete the file before restarting the server. Alternatively, before restarting the server, go to the Administrative Console -> Servers -> Application Servers -> <servername> -> IBM Service Logs. Remove the check from "Enable Service log" and save to the master configuration.

3. See the previous explanation on the function of the activity.log file. A node agent and all of the server under share the same activity.log file. In large environments where there a number of servers exists, some contention exists for writing messages to the activity.log file.

When the configured size of the file differs between the various servers which access the same activity.log file, the file is resized each time a message is written to the log, This could take a considerable amount of time and cause the server to appear to be slow. Messages are written to the activity.log file asynchronously (prior to WebSphere 6.0.2) so the server has to wait until the message is written and the activity.log.lck file (see above) .
To remedy this problem, go to Administrative Console -> Logs and traces -> <servername> -> IBM Service logs and make "Maximum log file" size specified the same value. This step has to be done for the nodeagent and servers that share the same activity.log file. For diagnostics, to eliminate this as a probable cause, remove the check from the "Enable Service log" in each of the servers and node agent.
 
 
Cross Reference information
Segment Product Component Platform Version Edition
Application Servers Runtimes for Java Technology Java SDK
 
 


Document Information


Product categories: Software > Application Servers > Distributed Application & Web Servers > WebSphere Application Server
Operating system(s): Windows
Software version: 6.0.2
Software edition:
Reference #: 1218842
IBM Group: Software Group
Modified date: Sep 30, 2005