Stability
Some solution solutions already implemented
- Sync stalling
- Rules about what is considered our best peer post-merge is not implemented
- Trying the same peer over and over again (shuffling the peers)
- Invalid block errors
- Consensus layer is on the fork with bad data
- Besu has a storage exception to report the invalid block to the consensus layer, which sets us off to a wrong fork (potential fix by Justin; GH issue)
- Potentially another Besu internal error could cause invalid blocks of which we don’t know about yet
...
- Worldstate root mismatch
- Bonsai and snapshots
- Solution: ??Confirmed working for many cases, needs more testing and handling of any edge cases
Issues around peering
- Need restart to find new peers sometimes during sync
- Potentially because of the lack of evaluating of peers during sync and post sync
- Solution: ??
- Losing many peers
- Because threads were blocked, we lost many peers
- Vertx for example uses different approaches to threading
- Solution: ??
...