benchmark
benchmark copied to clipboard
Produce-only requests should be able to use all worker nodes when using distributed workers
Currently if you specify a "no consume" workload (e.g., with 0 subscriptions per topic) it will still allocate 1/2 (rounded up) of all worker nodes to consumer workers, even though consumers will not be started.
Instead, we should allocate all nodes to producers.
Yes, it does make a lot of sense to make that improvement. The current logic, was to try to keep producer/consumers separated to make it easier to implement, although it's not really the best strategy.
This issue has been automatically marked as stale because it has not had recent activity. It will be closed after 5 days if no further activity occurs. Thank you for your contributions.