[
https://issues.apache.org/jira/browse/KAFKA-1397?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15145356#comment-15145356
]
Andrew Duch commented on KAFKA-1397:
------------------------------------
I would just like to add that I am still seeing issues deleting topics in
0.9.0. I will delete topics with kafka-topics.sh and it is set as "Marked for
Deletion" but never gets deleted. Is this still a known issue?
On server.log I see lot's of these messages:
[2016-02-12 21:33:29,480] WARN [Replica Manager on Broker 1]: While recording
the replica LEO, the partition [spiderman-pricegrabber-prices,5] hasn't been
created. (kafka.server.ReplicaManager)
On state-change.log, these are samples of the llog entries for this topic after
I tried to delete with kafka-topics.sh. For each of these entries, there is
typically one per partition... I've just summarized.
[2016-02-12 21:02:19,585] TRACE Controller 1 epoch 234 changed state of replica
2 for partition [spiderman-pricegrabber-prices,3] from
ReplicaDeletionIneligible to OfflineReplica (state.change.logger)
[2016-02-12 21:02:19,766] TRACE Controller 1 epoch 234 sending UpdateMetadata
request (Leader:-2,ISR:1,3,LeaderEpoch:0,ControllerEpoch:234) to broker 1 for
partition spiderman-pricegrabber-prices-5 (state.change.logger)
[2016-02-12 21:02:19,824] ERROR Controller 1 epoch 234 initiated state change
of replica 2 for partition [spiderman-pricegrabber-prices,6] from
OfflineReplica to ReplicaDeletionIneligible failed (state.change.logger)
java.lang.AssertionError: assertion failed: Replica
[Topic=spiderman-pricegrabber-prices,Partition=6,Replica=2] should be in the
ReplicaDeletionStarted states before moving to ReplicaDeletionIneligible state.
Instead it is in OfflineReplica state
[2016-02-12 21:02:20,117] TRACE Controller 1 epoch 234 changed state of replica
1 for partition [spiderman-pricegrabber-prices,4] from OfflineReplica to
ReplicaDeletionStarted (state.change.logger)
[2016-02-12 21:02:20,131] TRACE Broker 1 deleted partition
[spiderman-pricegrabber-prices,3] from metadata cache in response to
UpdateMetadata request sent by controller 1 epoch 234 with correlation id 1013
(state.change.logger)
[2016-02-12 21:02:20,132] TRACE Controller 1 epoch 234 received response
{error_code=0,partitions=[{topic=spiderman-pricegrabber-prices,partition=0,error_code=0}]}
for a request sent to broker Node(3, 10.108.0.105, 9092) (state.change.logger)
[2016-02-12 21:02:20,139] TRACE Broker 1 handling stop replica (delete=false)
for partition [spiderman-pricegrabber-prices,5] (state.change.logger)
[2016-02-12 21:02:20,139] TRACE Broker 1 finished handling stop replica
(delete=false) for partition [spiderman-pricegrabber-prices,5]
(state.change.logger)
[2016-02-12 21:02:20,140] TRACE Controller 1 epoch 234 received response
{error_code=0,partitions=[{topic=spiderman-pricegrabber-prices,partition=7,error_code=0}]}
for a request sent to broker Node(3, 10.108.0.105, 9092) (state.change.logger)
[2016-02-12 21:02:20,151] TRACE Broker 1 handling stop replica (delete=false)
for partition [spiderman-pricegrabber-prices,3] (state.change.logger)
[2016-02-12 21:02:20,151] TRACE Broker 1 finished handling stop replica
(delete=false) for partition [spiderman-pricegrabber-prices,3]
(state.change.logger)
[2016-02-12 21:02:20,152] TRACE Controller 1 epoch 234 received response
{error_code=0,partitions=[{topic=spiderman-pricegrabber-prices,partition=3,error_code=0}]}
for a request sent to broker Node(1, 10.108.0.122, 9092) (state.change.logger)
[2016-02-12 21:02:20,153] TRACE Broker 1 handling stop replica (delete=false)
for partition [spiderman-pricegrabber-prices,1] (state.change.logger)
[2016-02-12 21:02:20,838] TRACE Controller 1 epoch 234 received response
{error_code=0,partitions=[{topic=spiderman-pricegrabber-prices,partition=4,error_code=0}]}
for a request sent to broker Node(1, 10.108.0.122, 9092) (state.change.logger)
[2016-02-12 21:02:20,838] TRACE Controller 1 epoch 234 changed state of replica
1 for partition [spiderman-pricegrabber-prices,4] from ReplicaDeletionStarted
to ReplicaDeletionSuccessful (state.change.logger)
[2016-02-12 21:02:21,090] TRACE Controller 1 epoch 234 received response
{error_code=0,partitions=[{topic=spiderman-pricegrabber-prices,partition=6,error_code=0}]}
for a request sent to broker Node(3, 10.108.0.105, 9092) (state.change.logger)
[2016-02-12 21:02:21,090] TRACE Controller 1 epoch 234 changed state of replica
3 for partition [spiderman-pricegrabber-prices,6] from ReplicaDeletionStarted
to ReplicaDeletionSuccessful (state.change.logger)
> delete topic is not working
> ----------------------------
>
> Key: KAFKA-1397
> URL: https://issues.apache.org/jira/browse/KAFKA-1397
> Project: Kafka
> Issue Type: Bug
> Affects Versions: 0.8.2.0
> Reporter: Jun Rao
> Assignee: Timothy Chen
> Fix For: 0.8.2.0
>
> Attachments: KAFKA-1397.patch, KAFKA-1397_2014-04-28_14:48:32.patch,
> KAFKA-1397_2014-04-28_17:08:49.patch, KAFKA-1397_2014-04-30_14:55:28.patch,
> KAFKA-1397_2014-05-01_15:53:57.patch, KAFKA-1397_2014-05-01_18:12:24.patch,
> KAFKA-1397_2014-05-02_13:38:02.patch, KAFKA-1397_2014-05-05_11:17:59.patch,
> KAFKA-1397_2014-05-05_14:00:29.patch
>
>
> All unit tests are disabled since they hang transiently (see details in
> KAFKA-1391).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
