> On June 20, 2014, 8:35 p.m., Neha Narkhede wrote: > > core/src/main/scala/kafka/controller/KafkaController.scala, line 878 > > <https://reviews.apache.org/r/22496/diff/1/?file=608034#file608034line878> > > > > On second thoughts, we also don't need this API. The controller knows > > it's epoch when it becomes the controller and it doesn't need to read it > > later.
It won't know its epoch ZookeeperLeaderElector calls KafkaController.onControllerFailover and it won't get its epoch and tries to incrementControllerEpoch and fails. If I remove the above api RollingBounceTest fails I moved the readControllerEpochFromZookeeper to onControllerFailover before incrementControllerEpoch. - Sriharsha ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/22496/#review46316 ----------------------------------------------------------- On June 12, 2014, 4:52 a.m., Sriharsha Chintalapani wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/22496/ > ----------------------------------------------------------- > > (Updated June 12, 2014, 4:52 a.m.) > > > Review request for kafka. > > > Bugs: KAFKA-1096 > https://issues.apache.org/jira/browse/KAFKA-1096 > > > Repository: kafka > > > Description > ------- > > KAFKA-1096. An old controller coming out of long GC could update its epoch to > the latest controller's epoch. > > > Diffs > ----- > > core/src/main/scala/kafka/controller/KafkaController.scala > 8af48ab500779d3d851d25050e1308f5e7b588a6 > > Diff: https://reviews.apache.org/r/22496/diff/ > > > Testing > ------- > > > Thanks, > > Sriharsha Chintalapani > >
