Replies: 1 comment 1 reply
-
Why must you re-process the messages in the region that experienced the outage? Assuming that the consumer groups in the other regions perform the same business logic, there shouldn't be a need to process them again. |
Beta Was this translation helpful? Give feedback.
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hi all.
I have the following configuration for a geo-cluster:
A producer can submit messages and all the consumers will receive the messages.
The problem I am experiencing is the following:
If a region has no running consumers. That region will not process any messages. Should the consumer group come back up. There are no messages to process as the other regions have acknowledged them and these acknowledgements have already been replicated to the local cluster.
I have thought about trying to track the message ids however, with a cluster, the message ids will go in a round robin configuration. How will I know what is the minimum message to be processed? If I was running just 1 subscriber, this would be easy to track and then have a receiver process rewind and then process all the messages up to that point.
Has anyone faced this issue and how did you resolve it?
Thanks
Beta Was this translation helpful? Give feedback.
All reactions