Currently, a Nebula node will replay all the entries it has seen since creating the raft state. On slower hardware with synchronous write it can take quite a long time to start serving signing requests. We should compact the database to only keep some entries, and make the compaction run with every X entries passing.
Currently, a Nebula node will replay all the entries it has seen since creating the raft state. On slower hardware with synchronous write it can take quite a long time to start serving signing requests. We should compact the database to only keep some entries, and make the compaction run with every X entries passing.