One of our customer is facing an issue with his Call Manager. About 1 times on 2/3 weeks the CUP is raising until the limit (100%) and we need to restart the Tomcat service.
Based on bug tool I found that bug ID: CSCue33973
High CPU and memory usage when AXL is enabled and running for a long time.
It was discovered that the ArrayList hostNameList in class com.cisco.www.servletRouters.AXLThrottler had 14,847 entries in it. The list contained the server ip addresses multiple times. This ArrayList is supposed to hold a list of servers active in the cluster.
Upon looking at the source code it was discovered that the logic to populate that list was wrong and not only would that list continue to grow without bound or until Tomcat was restarted, but every 5 seconds the AXL Throttler will iterate that list and try to connect to each server in that list and query the database to check the dblcnqueue count. Not only was this happening 14,847 times every five seconds, that code block is synchronized and will block until it is done executing.
The longer between Tomcat restarts, the worse the problem is.
Someone can say me how I can check the list mentionned in the bug detail to see if we are facing that bug or not ?
I can only tell you that 8.6(2) is included in the Verified Releases, however after checking the release notes i do not see the bug stated: