As part of testing a new Exchange 2013 DAG before going into production, I was simulating different scenarios including server failure. One of the 3 servers in the DAG was powered off. Active databases failed over automatically to the other servers, and the databases mounted automatically.
When testing was complete, the server was powered back on. Running through my checks before moving onto the next server, I noticed that while both active and passive database copies where showing as healthy, the CopyQueueLength was 9223372036854773269
I found this blog that explained why this might happen, but didn’t give me the exact solution
In my case, I restarted the Microsoft Exchange Replication service on the affected server and the copy queue length returned to normal