Galileo runs in the Penumbra discord server and dispenses tokens to any Penumbra address which is posted in any channel to which it has the send-messages permission.
It does not duplicate command-line wallet management; rather, it shares a wallet by default with the
location of the wallet managed by the pcli command line Penumbra wallet. To set up Galileo, first
create a wallet with pcli, then send some tokens to that wallet on the test network. Then, you can
run Galileo:
You must clone the penumbra repo side-by-side with the Galileo repo, so that the dependencies are available as a relative path. This is a temporary workaround to support Git LFS in the Penumbra dependencies. See GH29 for details.
RUST_LOG=galileo=info DISCORD_TOKEN=<YOUR DISCORD TOKEN HERE> cargo run --release serve 100penumbra 1000test_usdThis will dispense 100 penumbra tokens and 1000 test_usd tokens to each address posted in any server to which the bot is
joined. Note that you must specify the bot's Discord API token using the DISCORD_TOKEN environment
variable in order to authenticate with Discord.
On first synchronization, the wallet must be caught up to speed with the state of the chain, which
can take some time; the info-level log output will inform you when the bot is ready. In the
meantime, it captures all the addresses which it observes, and holds them in memory until it's ready
to dispense tokens after completing first synchronization.
A variety of options are available, including adjusting rate-limiting, synchronization and
checkpointing intervals, and changing which node to connect to (the default is the hosted Penumbra
default testnet). Use the --help option for more details.
Users of the testnet can post a wallet address to the #testnet-faucet channel, and Galileo will
will give them a few funds. We ratelimit those requests to once per day per Discord user.
Since we destroy chain state between testnets (as of 2022Q4), users would lose funds if we don't
manually carry over allocations between testnets. Galileo can help with that!
The following steps assume that the penumbra repo and the galileo repo are sitting side-by-side in the filesystem. To update the allocations:
# Switch to galileo repo, run history crawl.
cd galileo/
cargo run --release -- history \
--channel https://discord.com/channels/824484045370818580/915710851917439060 \
> ../penumbra/testnets/discord_history.csv`
# Switch to penumbra repo.
cd ../penumbra/testnets/
# Generate new testnet directory, including most recent allocations scraped from Discord.
./new-testnet.sh
# Commit, push, and open a PR into the 'penumbra' repo.
We perform these updates on best-effort basis, so it's OK if the updates lag somewhat. Ideally we'd automate the process (#17), but until then, we'll perform it manually, say, every other testnet.
When we deploy a new testnet, we must bounce Galileo to keep the faucet working. The steps are:
- Rebuild the Galileo container via GHA,
passing in the Penumbra tag version to build from, e.g.
v0.58.0. - Wait for the container build to complete, then run:
kubectl set image deployments -l app.kubernetes.io/instance=galileo galileo=penumbra-v0.58.0substituting the correct version in the tag name.
Eventually we should automate these steps so they're performed automatically as part of a release.
If you need to use the --catch-up flag, to dispense to prior messages in the Discord channel,
consider running that action locally as a one-off. We don't want to set the catch-up CLI arg
on the primary deployment, where service restarts will cause the catch-up logic to run again.
