Replicated leveDB infinite loop with failover

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

Replicated leveDB infinite loop with failover

khandelwalanuj
Hi,

Using ActiveMQ 5.10 and replicated leveldb.

Test Case:
1) Running 3 ActiveMQ brokers with zookeeper and using replicated leveldb store.
2) Stop one of the slave broker.
3) Delete the levelDB index and other files.
4) Restart the slave broker.

At this time I got this error in the slave broker:
java.io.IOException: invalid record position 314575339 (file: /u/khandl/ldb/000000000c80071d.log,.....)

and primary broker goes into infinite loop.

I saw same issue reported  https://issues.apache.org/jira/browse/AMQ-5300 and http://activemq.2283324.n4.nabble.com/replicatedLevelDB-errors-after-failover-td4674550.html 

Is this a bug or some workaround can be done ?


Thanks,
Anuj