George Danezis

We add a mechanism to estimate the average delay in consensus (using a weighted average with weight 1/2 is the last sequenced message / timeout) and then adapt the time...

[core / node-sync] Get semaphore permit before launching task

1

Previously we got items out of a (bounded) channel, launched a task, and then tried to get a semaphore permit. This means that in practice the channel was unbounded, and...

[core] ConsensusAdapter / ConsensusListener is not complete

2

The logic in ConsensusAdapter submit is not complete. Specifically, we submit a transaction after we check that it has not already been sequenced / locks are allocated. The `submit` logic...

Replace consensus number notification Atomic with async watch

1

Refactor the cleanup code to use a watch that notifies other components of the consensus number change, instead of actively polling a shared Atomic u64 at every iteration of each...

Added a control system to limit the emission of headers under low volume

2

This PR is a quick one as I am re-learning the N+T codebase. What is does: - It records the number of batches (own + other) seen by the primary....

Receiver logic is vector for network DoS

4

Our ETH colleagues identified the uncontrolled creation of tasks, upon TCP connection, as well as the lack of any rate limiting / fairness / etc as a vector for DoS...

Ok, so we have now upgraded to support OpenSSL 1.1 API, but in the process the support for OpenSSL 1.0 was broken. Have a look at the failing Travis test,...

https://github.com/gdanezis/pybft/blob/131ecea621c1866b802af47f6901c7833688ad47/pybft/replica.py#L131-L141 In order to make progress (liveness) I think it is necessary for primaries to re-send a preprepare, send before, on a new request. This is however not in the...

* Added profile module * Annotated task launch to use the profiler * Added async_profile script to make pretty graphs The profiler requires the package `graphviz` and is used as...