handle component failure-> if we dump out mid-block processing we need to make sure our components will pick up the new block when restarted. Can make use of either a queueing system, or shared memory here (shm owned by the operator maybe?)
side notes:
rocksdb checkpoint is basically a 'fork' of the db...
before writing to a checkpoint 'fork', we need to make sure another 'fork' of the checkpoint exists,
in case we need to roll back AGAIN
handle component failure-> if we dump out mid-block processing we need to make sure our components will pick up the new block when restarted. Can make use of either a queueing system, or shared memory here (shm owned by the operator maybe?)
side notes: