Research reference implementation of the Alpenglow consensus protocol.
A local cluster example can be run with the following command:
./run.shThis spawns a local cluster with 6 nodes.
These nodes communicate via UDP on localhost.
Console output from the fastrace crate shows the progress of all nodes.
Further, we provide the simulations binary target.
It provides various simulations of parts of the Alpenglow protocol,
specifically resilience of Rotor, as well as latency and bandwidth requirements of Alpenglow as a whole.
Currently, configuration is a bit cumbersome and is done via the const values in src/bin/simulations/main.rs.
Some of the simulations, specifically the latency simulation, requires first downloading a public ping dataset via this script:
./download_data.shAfter that, the simulations can be run like this:
RUST_LOG="simulations=debug" cargo run --release --bin=simulationsSome micro-benchmarks can be run like this:
cargo benchRegular tests can be run like this:
./test.shThe more extensive test suite, including some slow tests, can be run like this:
./test.sh slowThere is a rudimentary implementation of a standalone node in the node binary. To use it, please do the folowing.
Since there is no gossip and stake manipulation in this prototype, you need to define all peers manually in advance. To do that, prepare a text file defining the socket addresses of the nodes that will be used in the test, e.g.
127.0.0.1:3000
127.0.0.1:3003
127.0.0.1:3006
127.0.0.1:3009
Obviously, you can use any IP addresses here, as long as they are reachable. Only first out of 3 needed ports is specified, the others are port+1 and port+2.
► cargo run --release --bin node -- --generate-config-files ip_list --config-name ag_node
will produce a config file starting with ag_node for each socket address found in ip_list file.
You may want to use multiple terminals or a script to start the nodes:
cargo run --release --bin node -- --config-name ag_node_0.toml &
cargo run --release --bin node -- --config-name ag_node_1.toml &
cargo run --release --bin node -- --config-name ag_node_2.toml &
cargo run --release --bin node -- --config-name ag_node_3.tomlYou should be able to take one node offline and bring it back up at will without cluster stopping.
For security related issues, please do not file a public issue on GitHub, instead reach out directly via email to:
quentin (at) anza (dot) xyz
Copyright (c) Anza Technology, Inc.
Licensed under the Apache License, Version 2.0. See LICENSE for details.