With PR #101 , I still run into some issues. When a consumer is looping 62s for the previous partition owner to release it, some new consumer may join, so the all of the assignment will be calculated again.
In this case, it make no sense to waiting for the partition calculated based-on the old consumer list. It should break the waiting loop and trying to claim all of the new assigned partitions.
We're experiencing the same issue, and the PR proposed here fixes it (thank you, @mrcold !)
@wvanbergen could you take a look at accepting this change, please? Thank you!
With PR #101 , I still run into some issues. When a consumer is looping 62s for the previous partition owner to release it, some new consumer may join, so the all of the assignment will be calculated again.
In this case, it make no sense to waiting for the partition calculated based-on the old consumer list. It should break the waiting loop and trying to claim all of the new assigned partitions.