asonnino
commented
11 months ago - Tweak a variety of system parameters to accommodate higher load
- Improve the synchroniser by waiting a bit before requesting missing blocks (in case they were just slow to arrive)
- Adapt the orchestrator to support crash-faults
- Do not wait for disconnected leaders
- Optionally use NVMe drives
- Parameterise consensus-only mode