Kafka Producer Metrics strange values

[Joao Reis <jo...@blip.pt>]

Hi guys,

We have a 3 producers producing for a kafka topic with 3 partitions.
We have a scheme were each producer, produces for one partition. All producers connect to all partitions but we ensure that each producer only sends messages for one partition.

The odd thing is that when looking to the producer metrics, kafka.producer:type=producer-node-metrics,client-id=producer-N,node-id=node-X (where N is producer number and X the leader node of the partition it connects to) for the client-id that should not have any message flowing, we get the following values:

incoming-byte-rate       876.0695011188626
outgoing-byte-rate       174.53683898779164
request-latency-avg       0.0
request-latency-max      -Infinity
request-rate                3.356477672842147
request-size-avg        52.0
request-size-max        52.0
response-rate                3.356588126892194


The request-latency-avg, request-latency-max, request-size-avg and request-size-max never change and its values seems to have never being set, but all the others are always changing. This looks like a bug in metrics and we have almost confidence that no messages are flowing through this client-id (which is what we want), but we need to know more why this strange behaviour in metrics.

All other clients, were messages are really flowing don’t have this issue and their metrics are fine.

Has anyone had this kind of issue ?

Thanks all for your help.

João Reis
Portugal



______________________________________________________________________
This email has been scanned by the Symantec Email Security.cloud service.
For more information please visit http://www.symanteccloud.com
______________________________________________________________________