1 Reply Latest reply on Feb 23, 2011 11:06 AM by Gary Tully

    KahaDBVersionTest test hangs due to corrupt data in activemq-core/src/test

    Dominic Tootell Newbie

      The unit test within activemq-core:  org.apache.activemq.store.kahadb.KahaDBVersionTest, within the source distribution bundle for fuse 5.4.2-fuse-01-00 hangs.

       

      With the source distribution of activemq from apache, this test takes around 10-15seconds to complete.

       

      When the test hangs you will see the following in activemq-core/target/activemq-test.log:

       

      -


      Dominic-Tootells-MacBook-Pro-2:target dominict$ more activemq-test.log

      2011-02-16 10:44:46,683 main           INFO  PListStore                     - PListStore:target/activemq-data/localhost/tmp_storage started

      2011-02-16 10:44:46,690 main           INFO  BrokerService                  - Using Persistence Adapter: KahaDBPersistenceAdaptertarget/activemq-data/kahadb/versionDB

      2011-02-16 10:44:46,730 main           INFO  Journal                        - Corrupt journal records found in 'target/activemq-data/kahadb/versionDB/db-1.log' between offsets: 72

       

      ....

      ....

       

      2011-02-16 10:44:47,181 main           INFO  MessageDatabase                - KahaDB is version 872415232

      2011-02-16 10:44:47,183 main           ERROR BrokerService                  - Failed to start ActiveMQ JMS Message Broker. Reason: java.io.EOFException

      java.io.EOFException

      at java.io.RandomAccessFile.readFully(RandomAccessFile.java:383)

      at java.io.RandomAccessFile.readFully(RandomAccessFile.java:361)

       

      ....

      ....

      2011-02-16 10:44:47,461 main           INFO  BrokerService                  - Using Persistence Adapter: KahaDBPersistenceAdaptertarget/activemq-data/kahadb/versionDB

      2011-02-16 10:44:47,461 main           INFO  MessageDatabase                - Database target/activemq-data/kahadb/versionDB/lock is locked... waiting 10 seconds for the database to be unlocked. Reason: java.io.IOException: File 'target/activemq-data/kahadb/versionDB/lock' could not be locked as lock is already held for this jvm.

      -


       

      This issue seems to stem from two kahadb data directories that are part of the source distribution within the activemq-core module:

       

      -


      ls -l src/test/resources/org/apache/activemq/store/kahadb/

       

      drwxr-xr-x@ 5 dominict  staff  170 15 Feb 21:21 KahaDBVersion1

      drwxr-xr-x@ 5 dominict  staff  170 15 Feb 21:21 KahaDBVersion2

      -


       

       

       

      Replacing these directories with those, that come with apache 5.4.2 activemq distribution, makes the issue go away; and the unit test run in the normal 10-15seconds.

       

      My question is:

       

       

      A) Is this "failing" data a representation of a some state that the kahadb persistence store can get into, and an activemq instance could fail with this issue?

       

      or

       

      B) Is this data just some bad kahadb data that would never occur in a live broker scenario?

       

      (Sorry I don't seem to be able to create a jira issue in the issue tracker for the above)

       

      Thanks in advance

      /dom