We had an Exchange 2007 CCR cluster where the passive node databases stopped seeding. After some investigation we noticed that the Exchange Replication Service starts and a second or two later stops. The event log did not give us much information, only stating that the service started and stopped again.
We checked all the databases on the passive node and noticed that they were in a dirty shut down state as the logs stopped replaying to the database a few days earlier.
Resolution that worked for us:
1. Move each database to a new folder on the disk that it resides.
2. Start the Exchange Replication Service. (We monitored the service and noted that it did not stop again.
3. Opened Exchange Management Console, chose a database and selected to update the database copy. Made sure to tick the option to delete the log files on the passive node.
4. Exchange started the seeding process and we deleted the copy of the database that was moved to the new folder to create disk space for the new seeded database.
5. Seeding finished and the copy status was healthy again for that database. We then continued with the other databases.