​Hello Bryan,

One more thing, I have just checked the *config/consumer.properties* where
the consumer timeout is not set
Does it have to do anything with this?

#consumer timeout
#consumer.timeout.ms=5000

# timeout in ms for connecting to zookeeper
zookeeper.connection.timeout.ms=6000​


Thank you!
​
______________________

*Kind Regards,*
*Anshuman Ghosh*
*Contact - +49 179 9090964*


On Tue, Mar 21, 2017 at 5:12 PM, Anshuman Ghosh <
[email protected]> wrote:

> Hello Bryan,
>
> Thank you so much for your reply!
>
> Just to answer your questions -
>
>    1. Back Pressure is set to the next processor but that isn't triggered
>    anytime (threshold limit is 1 GB)
>    2. No other custom properties are set apart from the one I have shared
>    with you.
>
> Also it suggests to increase the *session timeout *value - *Where do we
> do that*?
> Settings for my Kafka cluster is* zookeeper.connection.timeout.ms
> <http://zookeeper.connection.timeout.ms>=6000*
>
>
> Thank you!
> ​
> ______________________
>
> *Kind Regards,*
> *Anshuman Ghosh*
> *Contact - +49 179 9090964 <+49%20179%209090964>*
>
>
> On Tue, Mar 21, 2017 at 3:36 PM, Bryan Bende <[email protected]> wrote:
>
>> Anshuman,
>>
>> The error message is saying that NiFi tried to use a consumer where
>> the session already timed-out. So for example, if the session timeout
>> was 10 seconds, and NiFi had a consumer that hadn't been used in 12
>> seconds and then tried to use it, you would get this error because the
>> session for that consumer already exceeded the 10 seconds and expired.
>>
>> The typical reasons you would see that error message are the following...
>>
>> - The processor run schedule exceeds the duration of the
>> session.timeout (i.e. processor scheduled to run every 30 seconds,
>> session.timeout is 10 seconds)
>> - The processor wasn't scheduled to run for some period of time due to
>> back-pressure, so even though the run schedule might be every 1
>> second, it actually didn't execute for longer than the session.timeout
>> - The amount of time to perform a single poll() from Kafka takes
>> longer than the session.timeout, this is unlikely for NiFi but
>> theoretically possible for other consumers depending how they are
>> implemented
>>
>> From your screenshots it looks like ConsumeKafka_0_10 is scheduled to
>> run every 2 seconds...
>>
>> Do you ever have back-pressure between ConsumeKafka and the next
>> processors?
>> Do you have any other custom properties set in ConsumeKafka? I can
>> only see down to Max Uncommitted Time
>>
>> The good news is that newer versions of the Kafka 0.10.x client have
>> some built in heart-beating such that even if a consumer hasn't been
>> used in a while, it will continue to keep the session alive.
>>
>> Koji actually has a PR up to address this:
>> https://github.com/apache/nifi/pull/1527
>>
>> Thanks,
>>
>> Bryan
>>
>>
>> On Tue, Mar 21, 2017 at 9:33 AM, Anshuman Ghosh
>> <[email protected]> wrote:
>> > Hello,
>> >
>> > Can anyone please help me with this error message "ConsumeKafka_0_10"
>> > processor.
>> > Earlier I was not getting these messages, now I am seeing this Error and
>> > Waring message (attached screenshot) and sometimes getting duplicates :(
>> > Processor is running though!
>> >
>> > I have attached the configuration parameters screenshots for both the
>> > Publisher and Consumer processor.
>> >
>> > Thank
>> > ing
>> > you
>> > in advance
>> > !
>> >
>> > ______________________
>> >
>> > Kind Regards,
>> > Anshuman Ghosh
>> > Contact - +49 179 9090964
>> >
>>
>
>

Reply via email to