[jira] [Commented] (NIFI-5593) ConsumeKafka_0_11 topic name variable is not evaluated correctly

["Bryan Bende (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/NIFI-5593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16614845#comment-16614845 ] 

Bryan Bende commented on NIFI-5593:
-----------------------------------

[~maythes] this is expected behavior because when the processor is started it creates a pool of consumers for the topics specified at the time the processor is started, so at that point it evaluates whatever expression is in the topic name property. Once the consumer pool is created, then it is used until the processor is stopped.

You would need a very specific processor for your use case that would have knowledge of how you are creating a topic per hour and check the time and re-initialize the consumer pool, essentially it is very specific to your use case and I'm not sure it is something that makes sense to be provided for general use.

> ConsumeKafka_0_11 topic name variable is not evaluated correctly
> ----------------------------------------------------------------
>
>                 Key: NIFI-5593
>                 URL: https://issues.apache.org/jira/browse/NIFI-5593
>             Project: Apache NiFi
>          Issue Type: Bug
>    Affects Versions: 1.5.0
>         Environment: RHEL 7.3
> JAVA Oracle 1.8.0_144
> NiFi 1.5.0
> Processor name "ConsumeKafka_0_11"
>            Reporter: Georgy
>            Priority: Major
>         Attachments: proc_prop.PNG, proc_sched.PNG, proc_sett.PNG
>
>
> Hi guys,
> Found that "Topic Name" specified as expression "dmp_order_${now() :minus(3600000) :format('HH') }" is not evaluated correctly in "ConsumeKafka_0_11" processor after next hour begins.
> It is evaluated as expected after processor start/restart only.
> Say, the processor is started at 15:05. Value of "Topic name" variable will be "dmp_order_14". It's correct. It will consume data from the topic and everything is ok.
> After next hour begins, value of "Topic name" is not evaluated as "dmp_order_15". I see that consumer still tries to consume data from topic with name "dmp_order_14".
> I tried both Scheduling strategies "Timer driven" and "CRON driven" but have no success. Seems, I have to restart processor every time I want to re-evalueate "Topic name" variable.
> When I filled CustomText field with now() expression in "GenerateFlowFile" processor I didn't have such issues. It worked as expected with "CRON driven" strategy.
> Can you say is it a bug or normal behaviour for ConsumeKafka processor? 
> And if it is a normal behaviour can you say is there any opportunity to change "Topic name" dynamically in running processor?



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