We were able to get the secondary database node to take over processing.
We’re monitoring that, and still evaluating root cause.