I found some logs like this before everything started to go wrong ... [2014-12-02 07:08:11,722] WARN Partition [test3,13] on broker 2: No checkpointed highwatermark is found for partition [test3,7] (kafka.cluster.Partition) [2014-12-02 07:08:11,722] WARN Partition [test3,7] on broker 2: No checkpointed highwatermark is found for partition [test3,7] (kafka.cluster.Partition) ....
and further down [2014-12-02 07:08:11,740] INFO Truncating log test3-13 to offset 0. (kafka.log.Log) [2014-12-02 07:08:11,740] INFO Truncating log test3-9 to offset 0. (kafka.log.Log) [2014-12-02 07:08:11,740] INFO Truncating log test3-6 to offset 0. (kafka.log.Log) [2014-12-02 07:08:11,740] INFO Truncating log test3-28 to offset 0. (kafka.log.Log) [2014-12-02 07:08:11,741] INFO Truncating log test3-27 to offset 0. (kafka.log.Log) [2014-12-02 07:08:11,741] INFO Truncating log test3-15 to offset 0. (kafka.log.Log) [2014-12-02 07:08:11,741] INFO Truncating log test3-7 to offset 0. (kafka.log.Log) after that the logs from the first post starts. I don't know if that helps or explains anything.... /svante