Since the dev pool stay in place even when not being used, if the primary pools are not reachable or otherwise have some connectivity issue, cgminer will automatically fail over to another pool. There is a chance in that scenario that it could make it to the dev pool.

However, when the primary pool recovers, cgminer should automatically switch back within 5 minutes of identifying the pool as responsive again (it has a 5 minute holdback timer to make sure things stabilize).

This scenario could be what you have seen. There should be clear indication in the logs if this were the case.

Regardless, please do let me know if you isolate things any more. I am definitely interested in ensuring everything acts as expected.

Thank you,

Jason