By looking at your connection parameters, I am almost sure that “Minimum Pool Size =1” is causing that problem. What happen is when you set up minimum pool size to one than even there is no connection between IS and database like in case database goes down then IS forcefully tries to keep one pool connection alive which makes the JDBC connection stale. By setting “Minimum Pool Size = 0” would solve that problem.
setting minimum pool size = 0 may fix the hang problem, but it may have very negative impact on the performance (since every time you need to open new connection.
I’d guess you have a firewall between your WM server and DB, and it will timeout the connection over certain period of time.
To avoid it’s timing out, you can create a ping service to run, say, every 5 minutes. this will prevent it timing out. for this case, you can keep the setting minpoolSize =1 (not more than 1, since the ping will only reset one connection).
HTH,
Tong
what is your polling frequency ?
If your wM IS and DB are located in two different geographies and your you do polling tooo aggressive( say 10 sec ) if may cause notification to get hang. In past I have faced a couple of issues with polling notification getting hanged because of polling across WAN if network is slow.
what does wM guru recommended in above scenario, lets say there is a network latency between IS and DB and their is a business need to integrate two system in real time (accepted time is 2-3 minutes). Should we use polling notification or batch process ? or let custom application create a flat file , put it the FTP and wM IS do file polling ?
diging further i came across similar cases reported on advantage. wM support recommended to upgarde JVM to later release of 1.4.0.09 for their customer having HP UNIX on their floors. Not sure which jvm you are using. I attaching a consolidated list of similar cases reported to wM. Hope it helps some one. Polling Notification Issue.doc (39.5 KB)
