Write ahead log recovery is slow for some situation as outlined in #4873. The changes in #4873 and #4874 should be back-ported to 2.1 to speed up log recovery. To make these changes in 2.1 it would be good run continuous ingest from accumulo-testing with agitation on a small cluster. This would help verify log recovery is working properly.