Hi Luke,
Thanks for the details... so from explanation above, it seems that both of
these scenarios, I won't be able to avoid duplicates processing, which is
main purpose that I was looking to achieve
scenario 1: consumer shuts down, and doesn't commit offsets of
already polled and processed batc
Hi
1. I was under impression, from documentation, that close method waits for
30 seconds to complete processing of any in-flight events and then commits
offsets of last poll. Isn't that true? what does timeout of 30 seconds mean?
-> 30 seconds timeout is to have a buffer for graceful closing, ex:
Thanks Luke..
1. I was under impression, from documentation, that close method waits for
30 seconds to complete processing of any in-flight events and then commits
offsets of last poll. Isn't that true? what does timeout of 30 seconds mean?
2. How does CoperativeStickyAssignor solve my problem wh
Hi Pushkar,
Here's the answer to your questions:
> 1. During scale-down operation, I am adding a shutdown hook to the Java
Runtime, and calling close on the consumer. As per kafka docs, close
provides 30 sec to commit current offsets if auto.commit is enabled: so, i
assume that it will process th
Hi All,
I am hosting kafka consumers inside microservice hosted as kubernetes pods,
3 consumers in a consumer group.
There is a requirement to add auto-scaling where there will be a single pod
which will be auto-scaled out or scaled-in based on the load on
microservice.
So, 1 pod can be scaled out