Secure cryptocurrency custody using MPC.
Obviously, we need multiple parties to demonstrate or use MPC. In order to simulate a network of parties on our local machines, we use Docker Compose. You can find more information on how to install Docker here.
Once Docker is installed, you can spin up the first node on your local Codis network by running docker compose up bootstrap.
Record the address the bootstrap node is listening at, as we will use it later. Now, close the bootstrap and run it again
in daemon mode by adding a -d flag to the end of the original command.
Start a cluster of peers by running CODIS_BOOTSTRAPS=<bootstrap_address> docker compose up peer. Again, take note of
the peer addresses (at least one) as they will be used later.
The number of peers that will start is defined by the number of deploy replicas specified in the docker-compose.yaml file.
Finally, in another terminal, create a client by running CODIS_BOOTSTRAPS=<bootstrap_address> CODIS_HOST=<peer_address> docker compose up client.
You should see that the client has connected to the host and a terminal prompt should appear.
🚨 The client terminal interface is currently not working through Docker. You can still send requests to the network by setting the necessary environment variables on the local machine and running the client outside of Docker (refer to CODIS-T-23). 🚨
At this point, you might be wondering what the bootstrap service is. Why did we run a bootstrap service first instead of a peer service or a client service? To understand this, it is first important to know that Codis is a peer-to-peer application (using libp2p) and that the peers can have roles that they serve within the network.
The bootstrap service starts a bootstrap node that is used to keep and share a running list of any peers that connect to it. In this way, joining peers can quickly learn about the network if they connect to a bootstrap.
The peer service starts a set of peers that can execute the Codis protocols. These peers communicate with each other to generate the keys or signatures that are requested.
Codis uses Viper to source configurations from flags, files, and environment variables. Because of this, it can become a bit confusing on how a node is being configured—especially when running through Docker. The most important thing to remember is the precedence order for configuration: (1) flags, (2) environment variables, then (3) config file.
For development outside of Docker, it's easiest to not set environment variables and instead edit the config.yaml file. For development with Docker, it is necessary to configure services using environment variables—which you can edit for each service within the docker-compose.yaml file.