We can't just crash as we do currently, because there is no way to 'uncommit' a committed entry without truncating the log manually.Therefore we will just crash again on restart.
Current proposal is to log and return an error when it happens and continue without affecting the state machine.
We can't just crash as we do currently, because there is no way to 'uncommit' a committed entry without truncating the log manually.Therefore we will just crash again on restart.
Current proposal is to log and return an error when it happens and continue without affecting the state machine.