chandra kasiraju created KAFKA-6613:
---------------------------------------
Summary: The controller shouldn't stop partition reassignment
after an exception is being thrown
Key: KAFKA-6613
URL: https://issues.apache.org/jira/browse/KAFKA-6613
Project: Kafka
Issue Type: Bug
Components: admin, config, controller, core
Affects Versions: 0.11.0.2
Reporter: chandra kasiraju
Fix For: 0.11.0.2
I issued a partition reassignment command . It created the following entries in
the zookeeper .
But the entry never gets deleted because the partition reassigment hangs gets
some exceptions in kafka logs . After that no matter how many hours the
movement of partitions to other brokers never happens .
*Path in Zookeeper*
get /admin/reassign_partitions
{"version":1,"partitions":[{"topic":"__consumer_offsets","partition":44,"replicas":([1003,1001,1004,1002]},\{"topic":"683ad5e0-3775-4adc-ab55-82fda0761ba9_newTopic9","partition":0,"replicas":[1003,1004,1001,1002]},\{"topic":"683ad5e0-3775-4adc-ab55-82fda0761ba9_newTopic1","partition":0,"replicas":[1003,1004,1001,1002]},\{"topic":"__CruiseControlMetrics","partition":0,"replicas":[1002,1001,1004,1003]},\{"topic":"b1c39c85-aee5-4ea0-90a1-9fc7eedc635b_topic","partition":0,"replicas":[1003,1004,1001,1002]},\{"topic":"88ec4bd5-e149-4c98-8e8e-952e86ba5fae_topic","partition":4,"replicas":[1002,1004,1003,1001]},\{"topic":"c8c56723-73a5-4a37-93bf-b8ecaf766429_topic","partition":4,"replicas":[1002,1003,1004,1001]},\{"topic":"683ad5e0-3775-4adc-ab55-82fda0761ba9_newTopic9","partition":4,"replicas":[1002,1004,1003,1001]},\{"topic":"b1c39c85-aee5-4ea0-90a1-9fc7eedc635b_topic","partition":4,"replicas":[1003,1001,1004,1002]},\{"topic":"9db0cad2-69f8-4e85-b663-cd3987bd90fe_topic","partition":0,"replicas":[1003,1001,1004]},\{"topic":"683ad5e0-3775-4adc-ab55-82fda0761ba9_topic","partition":1,"replicas":[1003,1004,1001,1002]}]}
cZxid = 0x5000052f8
ctime = Tue Mar 06 01:27:54 UTC 2018
mZxid = 0x500005359
mtime = Tue Mar 06 01:28:06 UTC 2018
pZxid = 0x5000052f8
cversion = 0
dataVersion = 13
aclVersion = 0
ephemeralOwner = 0x0
dataLength = 1114
numChildren = 0
*Exception*
ERROR [KafkaApi-1002] Error when handling request
\{replica_id=1005,max_wait_time=500,min_bytes=1,max_bytes=10485760,isolation_level=0,topics=[{topic=__consumer_offsets,partitions=[{partition=41,fetch_offset=0,log_start_offset=0,max_bytes=1048576}]}]}
(kafka.server.KafkaApis)
kafka.common.NotAssignedReplicaException: Leader 1002 failed to record follower
1005's position 0 since the replica is not recognized to be one of the assigned
replicas 1001,1002,1004 for partition __consumer_offsets-41.
at kafka.cluster.Partition.updateReplicaLogReadResult(Partition.scala:274)
at
kafka.server.ReplicaManager$$anonfun$updateFollowerLogReadResults$2.apply(ReplicaManager.scala:1092)
at
kafka.server.ReplicaManager$$anonfun$updateFollowerLogReadResults$2.apply(ReplicaManager.scala:1089)
at
scala.collection.mutable.ResizableArray$class.foreach(ResizableArray.scala:59)
at scala.collection.mutable.ArrayBuffer.foreach(ArrayBuffer.scala:48)
at
kafka.server.ReplicaManager.updateFollowerLogReadResults(ReplicaManager.scala:1089)
at kafka.server.ReplicaManager.fetchMessages(ReplicaManager.scala:623)
at kafka.server.KafkaApis.handleFetchRequest(KafkaApis.scala:606)
at kafka.server.KafkaApis.handle(KafkaApis.scala:98)
at kafka.server.KafkaRequestHandler.run(KafkaRequestHandler.scala:66)
at java.lang.Thread.run(Thread.java:745)
I was expecting it would be recover from that exception move the partitions to
other nodes and finally remove the entries in /admin/reassign_partitions after
the move has happened.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
