mwtian
mwtian
The use case is indeed what "tracing" solutions address, but for Rust `tracing` I didn't find a way to use the span tracer on heap without guard, or connect spans...
This is an incomplete draft that broadcasts a request to sync digests. Remaining issues: - balance the need to start fetching certificate by digest as early as possible (e.g. during...
> Are we sure this doesn't break liveness? What happens if the sender node receives f+1 replies where f malicious replies come from Byzantine nodes and 1 honest reply comes...
> I saw your later comment about fetching the certificates instead of the CertificateDigests. Could you please provide more info on the desired approach here? We are following the approach...
Great to see more visibility here!
> Are we sure the header's author will re-try to retransmit its header if it needs more votes? Otherwise we may loose liveness (as you say) I need to read...
> > > Are we sure the header's author will re-try to retransmit its header if it needs more votes? Otherwise we may loose liveness (as you say) > >...
@akichidis I took a look at the code. Yes you are right that `ReliableNetwork` retries errors including timeouts, which I did not realize earlier. Indeed ack'ing Header does not yet...
> We should not put in a condition on the header round, even if it takes a lot of time to catch up and vote for it. In the case...
I think we should not do this for shared object certificate, since this does not take shared locks correctly. For owned object transactions, we don't have to queue the certificate...