Open
Description
Describe the bug
If max_workflow_task_polls
is 2 and workflow cache is enabled, it is expected that the worker will allocate one WFT poller to the normal task queue, and another to the sticky task queue.
Since Python 1.12.0, this is not happening. Network captures show that the Worker is only polling from the normal task queue, which is inefficient and increases load on the server side.