DB2, ReplayingCluster, and JdbcEventStore - "result set is closed"

[Peter Davis]

Just wondering if anyone has successfully used this combination: DB2, ReplayingCluster, and JdbcEventStore?  I am getting "Invalid operation: result set is closed" errors.

ReplayingCluster calls transactionManager.commitTransaction() after commitThreshold events. DB2 ResultSet does not have the HOLD_CURSORS_OVER_COMMIT flag set by default.  Unlike JpaEventStore, JdbcEventStore's Iterator implementation doesn't pre-fetch all of the events in a batch.  It also calls ResultSet.next() preemptively, instead of waiting until the next loop calls hasNext().  So it can't deal with the ResultSet being closed prematurely.

So far I haven't found how to set HOLD_CURSORS_OVER_COMMIT for these individual queries and it seems like a bad idea anyway.  (You can use Connection.setHoldability(HOLD_CURSORS_OVER_COMMIT) but that affects the whole pool.)

For my immediate use case we can get away with a NoTransactionManager, so that's what I've done for now.

Another option might be to create a second JdbcEventStore instance with a different DataSource for use by the ReplayingCluster -- no need for the EventStoreManagement to be transactional, just the event handlers.

org.axonframework.eventstore.EventStoreException: Exception occurred while attempting to fetch data from ResultSet

at org.axonframework.eventstore.jdbc.DefaultEventEntryStore$ResultSetIterator.hasNext(DefaultEventEntryStore.java:481) ~[axon-core-2.4.1.jar:2.4.1]

at org.axonframework.eventstore.jdbc.DefaultEventEntryStore$PreparedStatementIterator.hasNext(DefaultEventEntryStore.java:442) ~[axon-core-2.4.1.jar:2.4.1]

at org.axonframework.eventstore.jdbc.DefaultEventEntryStore$FilteredBatchingIterator.next(DefaultEventEntryStore.java:402) ~[axon-core-2.4.1.jar:2.4.1]

at org.axonframework.eventstore.jdbc.DefaultEventEntryStore$FilteredBatchingIterator.next(DefaultEventEntryStore.java:311) ~[axon-core-2.4.1.jar:2.4.1]

at org.axonframework.eventstore.jdbc.DefaultEventEntryStore$ConnectionResourceManagingIterator.next(DefaultEventEntryStore.java:293) ~[axon-core-2.4.1.jar:2.4.1]

at org.axonframework.eventstore.jdbc.DefaultEventEntryStore$ConnectionResourceManagingIterator.next(DefaultEventEntryStore.java:275) ~[axon-core-2.4.1.jar:2.4.1]

at org.axonframework.eventstore.jdbc.JdbcEventStore$IteratorDomainEventStream.initializeNextItem(JdbcEventStore.java:391) ~[axon-core-2.4.1.jar:2.4.1]

at org.axonframework.eventstore.jdbc.JdbcEventStore$IteratorDomainEventStream.next(JdbcEventStore.java:385) ~[axon-core-2.4.1.jar:2.4.1]

at org.axonframework.eventstore.jdbc.JdbcEventStore.doVisitEvents(JdbcEventStore.java:296) ~[axon-core-2.4.1.jar:2.4.1]

at org.axonframework.eventstore.jdbc.JdbcEventStore.visitEvents(JdbcEventStore.java:280) ~[axon-core-2.4.1.jar:2.4.1]

at org.axonframework.eventhandling.replay.ReplayingCluster$ReplayEventsTask.run(ReplayingCluster.java:303) ~[axon-core-2.4.1.jar:2.4.1]

at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511) [na:1.8.0_25]

at java.util.concurrent.FutureTask.run(FutureTask.java:266) [na:1.8.0_25]

at org.axonframework.common.DirectExecutor.execute(DirectExecutor.java:42) [axon-core-2.4.1.jar:2.4.1]

at org.axonframework.eventhandling.replay.ReplayingCluster.startReplay(ReplayingCluster.java:172) [axon-core-2.4.1.jar:2.4.1]

at org.axonframework.eventhandling.replay.ReplayingCluster.startReplay(ReplayingCluster.java:136) [axon-core-2.4.1.jar:2.4.1]

...

Caused by: com.ibm.db2.jcc.am.SqlException: [jcc][t4][10120][10898][4.19.26] Invalid operation: result set is closed. ERRORCODE=-4470, SQLSTATE=null

at com.ibm.db2.jcc.am.kd.a(Unknown Source) ~[db2jcc-4.19.26.jar:na]

at com.ibm.db2.jcc.am.kd.a(Unknown Source) ~[db2jcc-4.19.26.jar:na]

at com.ibm.db2.jcc.am.kd.a(Unknown Source) ~[db2jcc-4.19.26.jar:na]

at com.ibm.db2.jcc.am.ResultSet.checkForClosedResultSet(Unknown Source) ~[db2jcc-4.19.26.jar:na]

at com.ibm.db2.jcc.am.ResultSet.nextX(Unknown Source) ~[db2jcc-4.19.26.jar:na]

at com.ibm.db2.jcc.am.ResultSet.next(Unknown Source) ~[db2jcc-4.19.26.jar:na]

at com.mchange.v2.c3p0.impl.NewProxyResultSet.next(NewProxyResultSet.java:2859) ~[c3p0-0.9.1.1.jar:0.9.1.1]

at org.axonframework.eventstore.jdbc.DefaultEventEntryStore$ResultSetIterator.establishNext(DefaultEventEntryStore.java:487) ~[axon-core-2.4.1.jar:2.4.1]

at org.axonframework.eventstore.jdbc.DefaultEventEntryStore$ResultSetIterator.hasNext(DefaultEventEntryStore.java:478) ~[axon-core-2.4.1.jar:2.4.1]

... 52 common frames omitted

-Peter

[Allard Buijze]

Hi Peter,

I think a timeout has occurred on this question :-)

I have been thinking about this one for a while, and think that the JdbcEventStore should check for the resultset status each time the next item is retrieved. If the resultset is closed, it should perform a new query. Perhaps it should also read ahead a bit, so that the batch size could be larger than the transaction size.

For your scenario, it would absolutely make sense to use a different connection the reading, because it doesn't need to be part of the transaction anyway.

Cheers,

Allard

--
You received this message because you are subscribed to the Google Groups "Axon Framework Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to axonframewor...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Peter Davis]

Revisiting this issue -- finally had need for a transactional event handler, and discovered it is not so simple just to use a different connection for the event store management.  Must use a DataSourceConnectionProvider instead of a SpringDataSourceConnectionProvider, lest Spring synchronize the EventStoreManagement's Connection with the ReplayingCluster's open transaction, which causes it to commit anyway!

P.S. If I had a dollar for every line of Spring transaction code I've stepped through...

-Peter

[Allard Buijze]

Glad you got it working. Sometimes 'magic' comes with undesired side-effects.

PS. I'd be rich too ;-)