[
https://issues.apache.org/jira/browse/KAFKA-790?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14542808#comment-14542808
]
saiprasad mishra edited comment on KAFKA-790 at 5/13/15 9:58 PM:
-----------------------------------------------------------------
[~nehanarkhede]
I know its kind of late as this issue was closed long time back.
Is there a plan to put some kind of auto partition recovery or some mechanism
apart from putting in the log about this message as a
WARN where the througput drops as some of the partitions become defunct
I just saw this error happening in our kafka storm pipeline when one of the
kafka cluster in a DC had network issues and the leader election went bad
meaning some of the partitions were assigned a leader while the leader did not
have that partition and did not recover at all.
I had to move the all the topic partitions from this particular broker to other
healthy brokers
[2015-05-12 15:10:22,767] WARN [KafkaApi-6] Fetch request with correlation id
406573 from client ReplicaFetcherThread-0-6 on partition [qarth-asset-stg,93]
failed due to Partition [qarth-asset-stg,93] doesn't exist on 6
(kafka.server.KafkaApis)
This happened on a cluster kafka 0.8.1 version.
was (Author: saimishra):
[~nehanarkhede]
I know its kind of later as this issue was closed long time back.
Is there a plan to put some kind of auto partition recovery or some mechanism
apart from putting in the log about this message as a
WARN where the througput drops as some of the partitions become defunct
I just saw this error happening in our kafka storm pipeline when one of the
kafka cluster in a DC had network issues and the leader election went bad
meaning some of the partitions were assigned a leader while the leader did not
have that partition and did not recover at all.
I had to move the all the topic partitions from this particular broker to other
healthy brokers
[2015-05-12 15:10:22,767] WARN [KafkaApi-6] Fetch request with correlation id
406573 from client ReplicaFetcherThread-0-6 on partition [qarth-asset-stg,93]
failed due to Partition [qarth-asset-stg,93] doesn't exist on 6
(kafka.server.KafkaApis)
This happened on a cluster kafka 0.8.1 version.
> Kafka server throws UnknownTopicOrPartitionException right after becoming
> leader for a new partition
> ----------------------------------------------------------------------------------------------------
>
> Key: KAFKA-790
> URL: https://issues.apache.org/jira/browse/KAFKA-790
> Project: Kafka
> Issue Type: Bug
> Components: replication
> Affects Versions: 0.8.0
> Reporter: Neha Narkhede
> Assignee: Neha Narkhede
> Priority: Blocker
> Labels: kafka-0.8, p1
>
> 2013/03/05 01:24:01.207 INFO [ReplicaManager] [kafka-request-handler-5]
> [kafka] [] Replica Manager on Broker 273: Completed leader and isr request
> Name: LeaderAndIsrRequest; Version: 0; CorrelationId: 15
> 38; ClientId: ; AckTimeoutMs: 1000 ms; ControllerEpoch: 3;
> PartitionStateInfo: (foo,7) ->
> PartitionStateInfo(LeaderIsrAndControllerEpoch({ "ISR":"273,271",
> "leader":"273", "leaderEpoch":"0" },3),2),(foo,4) ->
> PartitionStateInfo(LeaderIsrAndControllerEpoch({ "ISR":"270,273",
> "leader":"270", "leaderEpoch":"0" },3),2),(foo,1) ->
> PartitionStateInfo(LeaderIsrAndControllerEpoch({ "ISR":"273,270",
> "leader":"273", "leaderEpoch":"0" },3),2); Leaders:
> id:270,host:host270,port:10251,id:273,host:host273
> ,port:10251
> 2013/03/05 01:24:01.473 INFO [Processor] [kafka-processor-10251-4] [kafka] []
> Closing socket connection to /172.20.72.66.
> 2013/03/05 01:24:01.505 INFO [Processor] [kafka-processor-10251-4] [kafka] []
> Closing socket connection to /172.20.72.244.
> 2013/03/05 01:24:01.521 INFO [Processor] [kafka-processor-10251-5] [kafka] []
> Closing socket connection to /172.20.72.243.
> 2013/03/05 01:24:01.555 INFO [Processor] [kafka-processor-10251-3] [kafka] []
> Closing socket connection to /172.20.72.11.
> 2013/03/05 01:24:01.568 INFO [Processor] [kafka-processor-10251-1] [kafka] []
> Closing socket connection to /172.20.72.66.
> 2013/03/05 01:24:01.606 INFO [Processor] [kafka-processor-10251-0] [kafka] []
> Closing socket connection to /172.20.72.66.
> 2013/03/05 01:24:01.758 INFO [Processor] [kafka-processor-10251-0] [kafka] []
> Closing socket connection to /172.20.72.66.
> 2013/03/05 01:24:01.776 ERROR [KafkaApis] [kafka-request-handler-7] [kafka]
> [] [KafkaApi-273] error when processing request (foo,1,0,1048576)
> kafka.common.UnknownTopicOrPartitionException: Topic foo partition 1 doesn't
> exist on 273
> at
> kafka.server.ReplicaManager.getLeaderReplicaIfLocal(ReplicaManager.scala:170)
> at
> kafka.server.KafkaApis.kafka$server$KafkaApis$$readMessageSet(KafkaApis.scala:301)
> at
> kafka.server.KafkaApis$$anonfun$kafka$server$KafkaApis$$readMessageSets$1.apply(KafkaApis.scala:268)
> at
> kafka.server.KafkaApis$$anonfun$kafka$server$KafkaApis$$readMessageSets$1.apply(KafkaApis.scala:264)
> at
> scala.collection.TraversableLike$$anonfun$map$1.apply(TraversableLike.scala:206)
> at
> scala.collection.TraversableLike$$anonfun$map$1.apply(TraversableLike.scala:206)
> at scala.collection.immutable.Map$Map2.foreach(Map.scala:127)
> at
> scala.collection.TraversableLike$class.map(TraversableLike.scala:206)
> at scala.collection.immutable.Map$Map2.map(Map.scala:110)
> at
> kafka.server.KafkaApis.kafka$server$KafkaApis$$readMessageSets(KafkaApis.scala:264)
> at kafka.server.KafkaApis.handleFetchRequest(KafkaApis.scala:231)
> at kafka.server.KafkaApis.handle(KafkaApis.scala:61)
> at kafka.server.KafkaRequestHandler.run(KafkaRequestHandler.scala:41)
> at java.lang.Thread.run(Thread.java:619)
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
