UpdDelAtomics failed with this stack trace:
ERROR [Network-Worker-Thread-8] 2012-08-23 17:25:58,506 ErrorResponse.java (line 91) ErrorCode of ERROR importance
java.lang.IllegalStateException: Record size 199 is too long for Transaction buffer in Transaction_100000016 depth=1 status=<ts=17,070,849 tc=UNCOMMITTED mvv=0> owner=Network-Worker-Thread-8
at com.persistit.Transaction.prepare(Transaction.java:1142)
at com.persistit.Transaction.writeStoreRecordToJournal(Transaction.java:1176)
at com.persistit.Transaction.store(Transaction.java:1105)
at com.persistit.Exchange.storeInternal(Exchange.java:1370)
at com.persistit.Exchange.store(Exchange.java:1294)
at com.persistit.Exchange.store(Exchange.java:2539)
at com.akiban.server.store.PersistitStore.writeRow(PersistitStore.java:376)
at com.akiban.server.store.PersistitStore.writeRow(PersistitStore.java:338)
at com.akiban.server.store.DelegatingStore.writeRow(DelegatingStore.java:80)
at com.akiban.qp.persistitadapter.OperatorStore.writeRow(OperatorStore.java:177)
at com.akiban.server.service.dxl.BasicDMLFunctions.writeRow(BasicDMLFunctions.java:583)
at com.akiban.server.service.dxl.HookableDMLFunctions.writeRow(HookableDMLFunctions.java:275)
at com.akiban.server.message.WriteRowRequest.executeDML(WriteRowRequest.java:69)
at com.akiban.server.message.WriteRowRequest.executeDML(WriteRowRequest.java:44)
at com.akiban.server.message.DMLRequest.execute(DMLRequest.java:86)
at com.akiban.server.service.network.DefaultRequestHandler.executeMessage(DefaultRequestHandler.java:78)
at com.akiban.server.service.network.DefaultRequestHandler.handleRequest(DefaultRequestHandler.java:58)
at com.akiban.server.service.network.RequestRunner.run(RequestRunner.java:72)
at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:886)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:908)
at java.lang.Thread.run(Thread.java:619)
I believe the circumstances are that our test automation software deleted the journal files, causing a prior journal I/O operation to fail with an IOException, specifically in the JournalManager#
In summary, I think this bug requires an extraordinary event (deleting the journal files of the running server), but the consequence also needs to be handled better.