I shut a single node instance down uncleanly. When I restarted it the logger did not have enough memory to preform the log sort, it got an OOME and died. I edited accumulo-env.sh and gave the logger process more memory. I restarted the logger process. However, the log recovery never restarted.
The master was continually printing message like the following.
After 20m I restarted the master and then log recovery proceeded.