In a large scale CFM fabric set up with very high fabric jobs, there could be stale device_job_mgr subprocess queues and job execution delays

Delay in creating job_status_consumer queue when the config push job has completed.

The scale of the fabric is related to the occurrence of the issue, increased number of devices in the fabric with frequent job submission.

There are many imprivments added in this regard on 21.4.L4. The final fix will be part of 21.4.L5