Rabu
22-06-2009, 15:59
Technology in use
I have a 3rd party tax application (vertex) that is running on Java 1.6 within Tomcat 6.0. The above is installed on VM servers (4GB of memory) with window server 2003 OS. For high availability reasons we are using the F5 load balancing software where we create a VIP that points to the 2 backend VM web servers running the window OS referenced previously. Tomcat Java Heap Memory allocation is 1024 MB for the initial and max memory pool.
The tax application has a series of web services (lookup tax, tax quote, invoice request) that our host ERP system calls fairly heavily throughout the day and quite intentsively throughout the nightly invoicing and sales update processes (200,000 - 300,000 web service calls within a 2 hour window).
Problem
We are experiencing application issues from our host system to the vertex tax application where tomcat does not seem to go down but the process stops unexpectedly. If tomcat went down I would see a log entry correct?
Action taken to date
we have turned on the Tomcat access logs to track when the application is being hit. We also monitoring this log for the F5 logic where the load balancer performs an http get against port 8080 to ensure that tomcat is up before sending incoming traffic to either web server
I have installed JConsole on both web servers and pushed test web server loads through and the boxes seem to handle the incoming traffic fine.
Help Needed
I have very limited knowledge with respect to JConsole application testing.
Does anyone have a good knowledge reference for application testing with JConsole. Such as looks at these memory variables or focus your attention on Garbarage Collection, etc.
What is a good Java memory heap allocation for an application that is called 200,000- 300,000 times within a 2 hours window. Obviously application design impacts memory but I guess I'm wondering if the 1024 max memory setting is too low? The fact that tomcat does not go down makes me think that 1024 could be fine?
I have a 3rd party tax application (vertex) that is running on Java 1.6 within Tomcat 6.0. The above is installed on VM servers (4GB of memory) with window server 2003 OS. For high availability reasons we are using the F5 load balancing software where we create a VIP that points to the 2 backend VM web servers running the window OS referenced previously. Tomcat Java Heap Memory allocation is 1024 MB for the initial and max memory pool.
The tax application has a series of web services (lookup tax, tax quote, invoice request) that our host ERP system calls fairly heavily throughout the day and quite intentsively throughout the nightly invoicing and sales update processes (200,000 - 300,000 web service calls within a 2 hour window).
Problem
We are experiencing application issues from our host system to the vertex tax application where tomcat does not seem to go down but the process stops unexpectedly. If tomcat went down I would see a log entry correct?
Action taken to date
we have turned on the Tomcat access logs to track when the application is being hit. We also monitoring this log for the F5 logic where the load balancer performs an http get against port 8080 to ensure that tomcat is up before sending incoming traffic to either web server
I have installed JConsole on both web servers and pushed test web server loads through and the boxes seem to handle the incoming traffic fine.
Help Needed
I have very limited knowledge with respect to JConsole application testing.
Does anyone have a good knowledge reference for application testing with JConsole. Such as looks at these memory variables or focus your attention on Garbarage Collection, etc.
What is a good Java memory heap allocation for an application that is called 200,000- 300,000 times within a 2 hours window. Obviously application design impacts memory but I guess I'm wondering if the 1024 max memory setting is too low? The fact that tomcat does not go down makes me think that 1024 could be fine?