[ 
https://issues.apache.org/jira/browse/NIFI-3498?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15872024#comment-15872024
 ] 

Joseph Witt commented on NIFI-3498:
-----------------------------------

[~peter_wilson] I feel like we've addressed this before but not seeing it right 
at the moment.  I would say that a great deal of refactoring and optimizing and 
improvements in general have gone into all things Kafka both consume and 
publish with NiFi in the 1.x line.

> PublishKafka processor incorrectly routing to success when it fails to write 
> to kafka
> -------------------------------------------------------------------------------------
>
>                 Key: NIFI-3498
>                 URL: https://issues.apache.org/jira/browse/NIFI-3498
>             Project: Apache NiFi
>          Issue Type: Bug
>    Affects Versions: 0.7.0
>            Reporter: Peter Wilson
>
> We were testing potential failure scenerios for our data ingestion process 
> which contains NiFi writing to Kafka using the PublishKafka processor. One of 
> our tests were to see how NiFi would respond if the Kafka brokers went down. 
> We have implemented a failure loop to ensure data does not get dropped if 
> they were unsuccessful in publishing. When we turned off the kafka broker, we 
> noticed the messages weren't being routed to failure and were instead being 
> treated as a successful message even though they weren't successful:
> e.g. log (Note, i've edited the ids):
> 2017-02-17 08:17:37,859 INFO [Timer-Driven Process Thread-4] 
> o.a.n.p.kafka.pubsub.PublishKafka PublishKafka[id=id] Successfully sent 
> StandardFlowFileRecord[uuid=uuid,claim=StandardContentClaim 
> [resourceClaim=StandardResourceClaim[id=id, container=default, section=19], 
> offset=157241, length=23661],offset=0,name=filename,size=23661] to Kafka as 0 
> message(s) in 30001 millis
> we would've expected this scenerio to be a failure. The NiFi processor's 
> Delivery Guarantee is set to "Guarantee Replicated Delivery" so we don't 
> expect to lose anything within the flow. Is there a fix for this?



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Reply via email to