[ 
https://issues.apache.org/jira/browse/GOBBLIN-19?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Kuai Yu reassigned GOBBLIN-19:
------------------------------

    Assignee: Kuai Yu
      Sprint: Apache Gobblin 170807

> dataset specific properties are ignored&dropped by KafkaBiLevelWorkUnitPacker
> -----------------------------------------------------------------------------
>
>                 Key: GOBBLIN-19
>                 URL: https://issues.apache.org/jira/browse/GOBBLIN-19
>             Project: Apache Gobblin
>          Issue Type: Bug
>            Reporter: Clemens Valiente
>            Assignee: Kuai Yu
>
> I failed to get dataset.specific.props to work on our jobs, and I think I 
> found the reason:
> in KafkaSource.getWorkUnitForTopicPartition the properties are added 
> correctly to the individual workunits.
> The KafkaBiLevelWorkUnitPacker then assigns the WorkUnits to their bins and 
> combines them into one WorkUnit in squeezeMultiWorkUnit() but doesn't copy 
> over the topicSpecificSettings.
> Using the KafkaSingleLevelWorkUnitPacker works fine with 
> dataset.specific.props since it doesn't call squeezeMultiWorkUnit on 
> non-empty workUnits.
>  
> *Github Url* : https://github.com/linkedin/gobblin/issues/1901 
> *Github Reporter* : [~cvaliente] 
> *Github Created At* : 2017-05-26T09:25:54Z 
> *Github Updated At* : 2017-05-31T06:39:04Z 
> h3. Comments 
> ----
> [~cvaliente] wrote on 2017-05-26T10:55:37Z : fix in #1903  
>  
> *Github Url* : 
> https://github.com/linkedin/gobblin/issues/1901#issuecomment-304253329 
> ----
> [~stakiar] wrote on 2017-05-30T17:42:07Z : Doesn't 
> `KafkaSource#addTopicSpecificPropsToWorkUnits` handle adding dataset specific 
> configuration? That method is run after the bin-packing is done. So if 
> `dataset.specific.props` isn't working I would guess the bug would be in that 
> method. 
>  
> *Github Url* : 
> https://github.com/linkedin/gobblin/issues/1901#issuecomment-304953996 
> ----
> [~cvaliente] wrote on 2017-05-31T06:39:04Z : You are right, that wasn't yet 
> implemented in 0.9 and I forgot to check upstream. 
>  
> *Github Url* : 
> https://github.com/linkedin/gobblin/issues/1901#issuecomment-305098396



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to