Closed JiaYingZhang closed 1 month ago
Hi @JiaYingZhang May I ask why did you close the PR?
@Gsantomaggio Because the producer must connect to the leader node with a load balancer, and my previous PR did not implement it.
A client application can still always connect to the load balancer and end up connected to the appropriate node. How can it do it? Two steps:
use the metadata command but ignore the information and always connect to the load balancer
retry connecting until it gets connected to the appropriate node
I will submit a new PR right away if it is still needed.
yes, please. We don't have space to follow this client, but it is on our plan to restart working on it. Any external contribution is welcome. Thank you
This PR add client workaound with a load balancer.
Client Workaround With a Load Balancer
The implementation method is similar to adding the load_malancer_mode option to the Python rstream client.