[adinn]

{quote}It sounds like you are saying node 2 is not involved in recovery at
all.{quote}
No, I was merely explaining how the TX manager and resource managers use persistent log records to
coordinate recovery. You can run recovery in each JVM, in which case the node which created a TX
recovers it. It is also possible to run recovery in a separate process in which case this process
recovers all transactions.
{quote}
I wonder then why these properties in arjuna have to be unique...
<property name="com.arjuna.ats.arjuna.xa.nodeIdentifier" value="1"/> in jbossts-properties.xml
and
jacorb.implname=JBoss in jacorb.properties
{quote}
The node identifier is used when creating XIDs, i.e. transaction identifiers. It has to be unique per
cluster node otherwise there is a high chance that two different JVMs in the cluster will try to use
the same XID for two different transactions. This would be a baaaaad thing.
If you are running recovery in each JVM (which is the default) then the recovery module uses the node
id to decide whether a recovered TX belongs to it or to some other node. You need to add each of the
node ids you allocated to your nodes into the list property xaRecoveryNodes to ensure that each node
does indeed recover its own transactions.
I am not sure what Jacorb does with jacorb.implname.

[azinemanas]

Are you using a XA datasource?

[deanhiller2000]

hmmm, no we are not.  We were just talking today about adding that as it might help some.  Here is some more data on the issue...

Only one app is showing this behavior when we pop the screen of that app.  Warnings ONLY show if BOTH nodes in the cluster are on.  If both nodes are not on, there is no warnings.  We will try the XAOracleDataSource and see if that fixes this problem.

thanks,

Dean

[deanhiller2000]

very weird, I backed out the jbossts-properties.xml file to just do the local tx and I still get the warnings in the logs with both nodes up.  I am guessing if I take one node offline, the warnings will go away like last time.  I am going to debug next to see at which point in our code indirectly causes these logs...

2009-12-17 15:19:26.213, '1420090918000000', 'OP1998', 116739)
2009-12-17 15:19:26,788 WARN  [jacorb.poa.controller] (RequestController-1) rid:
124 opname: saveRecord cannot process request, because object doesn't exist
2009-12-17 15:19:26,790 WARN  [jacorb.poa.controller] (RequestController-1) rid:
124 opname: saveRecord request rejected with exception:
2009-12-17 15:19:26,793 WARN  [com.arjuna.ats.jts.logging.loggerI18N] (http-0.0.
0.0-8080-1) [com.arjuna.ats.internal.jts.resources.errsavefail] ExtendedResource
Record.doSave failed. Returning default value: true
2009-12-17 15:19:26,818 WARN  [jacorb.poa.controller] (RequestController-1) rid:
126 opname: commit cannot process request, because object doesn't exist
2009-12-17 15:19:26,818 WARN  [jacorb.poa.controller] (RequestController-1) rid:
126 opname: commit request rejected with exception:
2009-12-17 15:19:26,972 INFO  [net.voicelog.customLiveOperator.handler.MainHandl
er] (http-0.0.0.0-8080-1) User: OP1998, POP TIME: 4714 ms

[deanhiller2000]

oh, there is more to backing out than just that one file .  I moved forward instead and installed the XADriver ds.xml file and

that fixed it!!!

Now, how do I view the sql for a transaction stuck in the ActionStore......

Transaction -3f5773b9:bd0d:4b2a9d7d:42 still in ActionStore