You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@kafka.apache.org by "ASF GitHub Bot (JIRA)" <ji...@apache.org> on 2017/06/10 23:34:18 UTC
[jira] [Commented] (KAFKA-5427) Transactional producer cannot find
coordinator when trying to abort transaction after error
[ https://issues.apache.org/jira/browse/KAFKA-5427?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16045748#comment-16045748 ]
ASF GitHub Bot commented on KAFKA-5427:
---------------------------------------
GitHub user hachikuji opened a pull request:
https://github.com/apache/kafka/pull/3297
KAFKA-5427: Transactional producer should allow FindCoordinator in error state
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/hachikuji/kafka KAFKA-5427
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/kafka/pull/3297.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #3297
----
commit 76ffa96261084474cde0db350b96d8e4df9cbc55
Author: Jason Gustafson <ja...@confluent.io>
Date: 2017-06-10T23:31:32Z
KAFKA-5427: Transactional producer should allow FindCoordinator in error state
----
> Transactional producer cannot find coordinator when trying to abort transaction after error
> -------------------------------------------------------------------------------------------
>
> Key: KAFKA-5427
> URL: https://issues.apache.org/jira/browse/KAFKA-5427
> Project: Kafka
> Issue Type: Sub-task
> Components: clients, core, producer
> Reporter: Jason Gustafson
> Assignee: Jason Gustafson
> Priority: Blocker
> Fix For: 0.11.0.0
>
>
> It can happen that we receive an abortable error while we are already aborting a transaction. In this case, we have an EndTxnRequest queued for sending when we transition to ABORTABLE_ERROR. It could be that we need to find the coordinator before sending this EndTxnRequest. The problem is that we will fail even the FindCoordinatorRequest because we are in an error state. This causes the following endless loop:
> {code}
> [2017-06-10 19:29:33,436] DEBUG [TransactionalId my-first-transactional-id] Enqueuing transactional request (type=FindCoordinatorRequest, coordinatorKey=my-fi
> rst-transactional-id, coordinatorType=TRANSACTION) (org.apache.kafka.clients.producer.internals.TransactionManager)
> [2017-06-10 19:29:33,436] DEBUG [TransactionalId my-first-transactional-id] Enqueuing transactional request (type=EndTxnRequest, transactionalId=my-first-tran
> sactional-id, producerId=1000, producerEpoch=0, result=ABORT) (org.apache.kafka.clients.producer.internals.TransactionManager)
> [2017-06-10 19:29:33,536] TRACE [TransactionalId my-first-transactional-id] Not sending transactional request (type=FindCoordinatorRequest, coordinatorKey=my-
> first-transactional-id, coordinatorType=TRANSACTION) because we are in an error state (org.apache.kafka.clients.producer.internals.TransactionManager)
> [2017-06-10 19:29:33,637] TRACE [TransactionalId my-first-transactional-id] Request (type=EndTxnRequest, transactionalId=my-first-transactional-id, producerId
> =1000, producerEpoch=0, result=ABORT) dequeued for sending (org.apache.kafka.clients.producer.internals.TransactionManager)
> [2017-06-10 19:29:33,637] DEBUG [TransactionalId my-first-transactional-id] Enqueuing transactional request (type=FindCoordinatorRequest, coordinatorKey=my-fi
> rst-transactional-id, coordinatorType=TRANSACTION) (org.apache.kafka.clients.producer.internals.TransactionManager)
> [2017-06-10 19:29:33,637] DEBUG [TransactionalId my-first-transactional-id] Enqueuing transactional request (type=EndTxnRequest, transactionalId=my-first-tran
> sactional-id, producerId=1000, producerEpoch=0, result=ABORT) (org.apache.kafka.clients.producer.internals.TransactionManager)
> [2017-06-10 19:29:33,737] TRACE [TransactionalId my-first-transactional-id] Not sending transactional request (type=FindCoordinatorRequest, coordinatorKey=my-
> first-transactional-id, coordinatorType=TRANSACTION) because we are in an error state (org.apache.kafka.clients.producer.internals.TransactionManager)
> [2017-06-10 19:29:33,837] TRACE [TransactionalId my-first-transactional-id] Request (type=EndTxnRequest, transactionalId=my-first-transactional-id, producerId
> =1000, producerEpoch=0, result=ABORT) dequeued for sending (org.apache.kafka.clients.producer.internals.TransactionManager)
> [2017-06-10 19:29:33,838] DEBUG [TransactionalId my-first-transactional-id] Enqueuing transactional request (type=FindCoordinatorRequest, coordinatorKey=my-fi
> rst-transactional-id, coordinatorType=TRANSACTION) (org.apache.kafka.clients.producer.internals.TransactionManager)
> [2017-06-10 19:29:33,838] DEBUG [TransactionalId my-first-transactional-id] Enqueuing transactional request (type=EndTxnRequest, transactionalId=my-first-tran
> sactional-id, producerId=1000, producerEpoch=0, result=ABORT) (org.apache.kafka.clients.producer.internals.TransactionManager)
> [2017-06-10 19:29:33,938] TRACE [TransactionalId my-first-transactional-id] Not sending transactional request (type=FindCoordinatorRequest, coordinatorKey=my-
> first-transactional-id, coordinatorType=TRANSACTION) because we are in an error state (org.apache.kafka.clients.producer.internals.TransactionManager)
> {code}
> A couple suggested improvements:
> 1. We should allow FindCoordinator requests regardless of the transaction state.
> 2. It is a bit confusing that we allow EndTxnRequest to be sent in both the ABORTABLE_ERROR and the ABORTING_TRANSACTION states. Perhaps we should only allow EndTxnRequest to be sent in ABORTING_TRANSACTION. If we hit an abortable error and we are already aborting, then we should just stay in ABORTING_TRANSACTION and perhaps log a warning.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)