[ 
https://issues.apache.org/jira/browse/STORM-2994?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16400333#comment-16400333
 ] 

RAbreu commented on STORM-2994:
-------------------------------

I've been injecting in the Kafka topic two types of messages
 * messages that are incorrect and won't be emitted by the Spout
 * messages that are correct and will be emitted by the Spout

Using storm-kafka-clients 1.1.0, even the messages that are emitted pile up and 
aren't commited

 

 
{code:java}
2018-03-15T12:22:25.880Z o.a.s.k.s.KafkaSpout [DEBUG] - Polled [0] records from 
Kafka. [1379] uncommitted offsets across all topic partitions

{code}

The incorrect messages aren't added to the total of uncommitted messages.

Using storm-kafka-clients 1.1.2, the behaviour is different.

I uploaded the same topology, bumping storm-kafka-clients to 1.2.1, picking up 
the previous lag of > 14000 and reducing it to ~3000.
It didn't go to 0 and stayed firmly in that level.
Once I started injecting correct messages, eventually all messages are 
committed (not sure at this point if it is when he commit interval elapses)

Question:

Aren't the tuples to which 




 

 

 

> KafkaSpout consumes messages but doesn't commit offsets
> -------------------------------------------------------
>
>                 Key: STORM-2994
>                 URL: https://issues.apache.org/jira/browse/STORM-2994
>             Project: Apache Storm
>          Issue Type: Bug
>          Components: storm-kafka-client
>    Affects Versions: 1.1.0, 1.1.2
>            Reporter: RAbreu
>            Priority: Major
>
> A topology that consumes from two different Kafka clusters: 0.10.1.1 and 
> 0.10.2.1.
> Spouts consuming from 0.10.2.1 have a low lag (and regularly commit offsets) 
> The Spout that consumes from 0.10.1.1 exhibits either:
> 1- Unknown lag
> 2- Lag that increments as the Spout reads messages from Kafka
>  
> In DEBUG, Offset manager logs: "topic-partition has NO offsets ready to be 
> committed", despite continuing to consume messages.
> Several configuration tweaks were tried, including setting maxRetries to 1, 
> in case messages with a lower offset were being retried (logs didn't show it, 
> though)
> offsetCommitPeriodMs was also  lowered to no avail.
> The only configuration that works is to have 
> ConsumerConfig.ENABLE_AUTO_COMMIT_CONFIG=true, but this is undesired   since 
> we lose processing guarantees.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to