- not having lock can cause "block number list" to become corrupt => crash
- also may result in 2 threads receiving same block no => possble cause of
Bug #13987 Cluster: Loss of data nodes can cause high CPU usage from ndb_mgmd