Transaction Timeout Warning

[nday_catalyst]

I've been having a problem on site for a while now and I need advice on how to solve it. I get the following warning logged (nothing else is logged):
2006-08-11 13:39:27,959 WARN [org.jboss.tm.TransactionImpl] Transaction TransactionImpl:XidImpl [FormatId=257, GlobalId=cpaidi05.homedepot.com//969207, BranchQual=] timed out. status=STATUS_ACTIVE

When this warning is logged, a series of 15-20 of them will be logged in succession. After that, the application stops responding. Basically, anything that requires database access doesn't work after this message is logged. In addition, the jmx-console does not come up after this either (if it matters, we've changed jboss to persist queues to the DB2 database instead of hypersonic).

We are using JBoss 3.2.7 on IBM JVM 1.4.2 accessing a DB2 database. We have 3 datasources configured. Interestingly, the problem always happens against a specific datasource. The other two datasources are unaffected by the problem.

I've seen some other posts with the same warning, but the response was that it was an application problem because jdbc resources aren't being closed. I followed up on that in my application by setting the "track-statements" attribute of the datasource to true in an attempt to try and identify any resources that might not be getting closed. This didn't produce anything in the logs though. I'm assuming this means that my application is releasing resources properly.

The questions I have are:
- I'm kind of at my wits end with this. What approach should I use to troubleshoot this problem? Is there a certain set of jboss classes that I could turn logging on for that would give me an indication as to what is causing the transactions to time out.
- Is my interpretation of the track-statements attribute correct?

If there is any other information that I can provide, please let me know!!

Thanks in advance,
Neal

[ben_ben]

We've been having something similar, using JBoss 4 on JDK 1.4 contacting an Ingres database.

The connections in the connection pool get used up as timed-out connections, but still get returned to the pool. Which means that it then tries to use the now-dead connections and falls over. By using the 'check-valid-connection-sql' tag in our datasource we can ensure the system continues to run, but this doesn't solve the original timeout problem.

Any advice gratefully accepted!

[michaelukpong]

I have a very similar problem. I have a queue which i send about 200 messages to at once. Jboss JMS sends about 30 (back to back) then stalls, minutes later I start getting "transaction timed out globalid=..."

After that nothing works, the messages dont get delivered, even after restarting jboss.

Am I flooding the queue?

How can this be sorted out? Any help will be greatly appreciated