A rust go utility to create a set of Libra validator nodes in a kubernetes namespace.
Prerequisites: kubectl must be set up to access the target cluster. eg. kubectl get nodes should return something. Go will be required and rustc and cargo are needed to work with Libra.
TL;DR Just run the Apply function here and it does everything and then you have your own Libra test net.
See: https://developers.libra.org/docs/run-local-network
See: https://developers.libra.org/blog/2019/10/29/a-guide-to-running-libra-validators
See: https://github.com/libra/libra ( and do a git clone https://github.com/libra/libra.git)
Step two is to git clone the libra project and cd to the libra directory and run docker/validator/build.sh and then docker/mint/build.sh
This takes a ~20 minutes each and I had to give docker more cpu and memory to not fail. We'll try to avoid repeating that too much.
Then I label and push them:
docker tag e0e17d4611e0 gcr.io/fair-theater-238820/libra_e2e
docker tag ec5a821668c4 gcr.io/fair-theater-238820/libra_mint
docker push gcr.io/fair-theater-238820/libra_e2e
docker push gcr.io/fair-theater-238820/libra_mint
Later they should have version numbers.
Or, you could skip this step use the ones at my gcr which are public.
Next run the Apply() function in apply.go. The way I do it is that I run the test (TestApply) since I have to do this over and over and over. The way I brought it up is to first get the stateful set running and keeping libra from running by uncommenting:
command: ["/bin/sh"]
args: ["-c", "while true; do sleep 10000; done"]
in libra-validator and running. Then we can kubectl exec -it libra-0 -- bash and have a look around. We can check that the volume is mounted at /opt/libra/data and that the binary is at /opt/libra/bin/libra-node. Note that we can also see a binary at /opt/libra/bin/config-builder which claims to be able to generate config for us.
At the root there is a file /docker-run.sh whose contents are:
#!/bin/sh
# Copyright (c) The Libra Core Contributors
# SPDX-License-Identifier: Apache-2.0
set -ex
cd /opt/libra/etc
echo "$NODE_CONFIG" > node.config.toml
echo "$SEED_PEERS" > seed_peers.config.toml
echo "$NETWORK_KEYPAIRS" > network_keypairs.config.toml
echo "$CONSENSUS_KEYPAIR" > consensus_keypair.config.toml
echo "$FULLNODE_KEYPAIRS" > fullnode_keypairs.config.toml
exec /opt/libra/bin/libra-node -f node.config.toml
See: https://github.com/libra/libra/tree/master/config "Generating a new TestNet"
The config-builder will make the config we need. One wrinkle is that we don't know the IP address of the pods at this time, libra will NOT take a name instead of a dotted quad, so I'll use 100.100.100.100 and 101.101.101.101 etc as addresses. We create the secret key and then all the configs for every node in a folder and then copy them all into a k8s ConfigMap which we can mount as a directory in the pod. Then write some bash to replace the addresses with the real addresses and then call /opt/libra/bin/libra-node -f node.config.toml
kubeclt get po will show:
NAME READY STATUS RESTARTS AGE
libra-0 1/1 Running 0 12m
libra-1 1/1 Running 0 11m
libra-2 1/1 Running 0 11m
which looks good.
Port forward a pod: kubectl port-forward libra-0 8000:8000
and then, in a different terminal window, cd to the libra project and
cargo run --bin cli -- -a localhost -p 8000 -m "/Users/awootton/Documents/workspace/libra-statefulset/tmp/mint.key"
and change the path to the mint. When you get the libra% prompt type
query tr 0 1 false
which will confirm the zeroth transaction (see https://github.com/mikeholenderski/libra-docker)
Also, try this: https://developers.libra.org/docs/my-first-transaction which also works.