Atomikos Forum

In our environment using Tomcat 6.0.20 we keep having the following recordsin tm.out:

10-06-22 07:42:59,007 [Atomikos:1] Coordinator 10.200.10.146.tm0001900233 entering state: HEUR_HAZARD
10-06-22 07:42:59,150 [Atomikos:0] (1.5) executing task: com.atomikos.icatch.imp.Propagator$PropagatorThread@d43b78

This is just a snippit of tm.out. In fact tm.out has 5 entries per second about the HEUR_HAZARD.

We noticed that this is caused by content in the tmlogXX.log.
However we don't know how to stop this. On some of our machines this tmlogXX.log already reached the 414000 number. So the tmlog414231.log lives on that machine and this is continuesly rising. Each 50-60 seconds this file is deleted and created again with a higher number.

This is all is happening while our tomcat server is in idle mode. We don't do anything!

What can we do to solve this problem.
We know already that we can set the property com.atomikos.icatch.enable_logging to 'false'. But this is not recommended by Atomikos.
What else can we do to make sure it will not continue on all our hosting servers and developer machines?

Thx in advance.

Hi,

This is most likely a JMS or JDBC connection that is going down before the end of a transaction... I would check that first.

HTH

"This is most likely a JMS or JDBC connection that is going down before the end of a transaction... I would check that first."

Yes it could be, but the only thing I have is an old tmlog.log file which is copied from another machine. And this file is not viewable since it looks like a binary format.

So I'm not able to see if there is a JMS/JDBC connection inside the tmlog.log file.

My guess is that in the old file you had pending transactions from a prior database connector.

You can use the JMX administration tool to inspect this. See http://www.atomikos.com/Documentation/JmxInstrumentation for details. If all else fails, this tool allows you to delete the problematic transaction from the logs, thereby keeping the logs from growing.

HTH