Transitioner problem solved
It turned out the problem was actually caused by a corrupted index in one of the primary database tables used by BOINC Server. Recreated the table/index and things are looking much better so I've restarted the system. However, transitioner is now going through all the 600k workunits in the DB and only once that is done, the system will start creating new workunits. Until then we are out of work to send. Sorry! I'll keep an eye out to make sure we recover fully.

More...