[ https://issues.apache.org/jira/browse/KAFKA-42?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13468947#comment-13468947 ]
Jun Rao commented on KAFKA-42: ------------------------------ Thanks for patch v2. Some more comments: 20. ReassignPartitionsCommand: 20.1 Could we add a description of the format of the jaon file in the command line option? 20.2 If partitionsToBeReassigned is an empty, should we just fail the command? 20.3 reassignPartitions(): Instead of check the existence of ReassignPartitionsPath and then write in ZK, it's better to use ZkUtils.createPersistentPath(), which throws an exception if node already exists. This will prevent the corner case that the path is created just after the existence check. 20.4 createReassignedPartitionsPathInZK: It seems that each call to this method just overwrites ReassignPartitionsPath with 1 partition's assignment. So we will lose the assignments of all partitions except the last one? 21. CheckReassignmentStatus: It's better to move checkIfReassignmentSucceeded and checkIfPartitionReassignmentSucceeded from ZkUtils to CheckReassignmentStatus since they are only used here and ZkUtils is getting big. 22. KafkaController.onBrokerStartup() : It seems that we can get partitionsBeingReassigned from the cache in controllerContext, instead of from ZK. 23. PartitionStateMachine.initializeLeaderAndIsrForPartiiton(): When writing the initial leaderAndIsr path for a new partition, there is no need to read the path first to make sure that it doesn't exists. createPersistentPath will throw an exception if the path exists. > Support rebalancing the partitions with replication > --------------------------------------------------- > > Key: KAFKA-42 > URL: https://issues.apache.org/jira/browse/KAFKA-42 > Project: Kafka > Issue Type: Bug > Components: core > Reporter: Jun Rao > Assignee: Neha Narkhede > Priority: Blocker > Labels: features > Fix For: 0.8 > > Attachments: kafka-42-v1.patch, kafka-42-v2.patch > > Original Estimate: 240h > Remaining Estimate: 240h > > As new brokers are added, we need to support moving partition replicas from > one set of brokers to another, online. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira