blockworks-foundation
/
pyth-crosschain
mirror of https://github.com/blockworks-foundation/pyth-crosschain.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
pyth-crosschain/ethereum/Deploying.md

8.0 KiB
Raw Blame History

Deploying Contracts to Production

Running the Truffle migrations in migrations/prod or migrations/prod-receiver will deploy the contracts to production. The prod-receiver migrations should be used when you need to deploy to a chain that is unsupported by the Wormhole network. The Wormhole Receiver contract acts as a read-only Wormhole endpoint that can verify Wormhole messages even if the Wormhole network has not yet connected the chain.

This is the deployment process:

# The Secret Recovery Phrase for our deployment account.
export MNEMONIC=...

# Deploy the changes
# You might need to repeat this script because of busy RPCs. Repeating would not cause any problem even
# if the changes are already made. Also, sometimes the gases are not adjusted and it will cause the tx to
# remain on the mempool for a long time (so there is no progress until timeout). Please update them with
# the network explorer gas tracker. Tips in Troubleshooting section below can help in case of any error.
./deploy.sh <network_a> <network_b> <...>
# Example: Deploying to some testnet networks
# ./deploy.sh bnb_testnet fantom_testnet mumbai
#
# Example: Deploying to some mainnet networks
# ./deploy.sh ethereum bnb avalanche

# Perform this in first time mainnet deployments with Wormhole Receiver. (Or when guardian sets are upgraded)
npm run receiver-submit-guardian-sets -- --network $MIGRATIONS_NETWORK

As a sanity check, it is recommended to deploy the migrations in migrations/prod to the Truffle development network first. You can do this by using the configuration values in .env.prod.development.

As a result of this process for some files (with the network id in their name) in networks and .openzeppelin directory might change which need to be committed (if they are result of a production deployment).

If you are deploying to a new network, please add the new contract address to consumer facing libraries and documentations.

To do so, add the contract address to both Pyth Gitbook EVM Page and pyth-evm-js package. You also need to add the new network address to pyth-evm-js relaying example.

networks directory

Truffle stores the address of the deployed contracts in the build artifacts, which can make local development difficult. We use truffle-deploy-registry to store the addresses separately from the artifacts, in the networks directory. When we need to perform operations on the deployed contracts, such as performing additional migrations, we can run npx apply-registry to populate the artifacts with the correct addresses.

Each file in the network directory is named after the network id and contains address of Migration contract and PythUpgradable contract (and Wormhole Receiver if we use prod-receiver). If you are upgrading the contract it should not change. In case you are deploying to a new network make sure to commit this file.

.openzeppelin directory

In order to handle upgrades safely this directory stores details of the contracts structure, such as implementation addresses and their respective storage layout in one file per network (the name contains network id). This allows truffle to check whether the upgrade is causing any memory collision. Please take a look at (this doc)[https://docs.openzeppelin.com/upgrades-plugins/1.x/writing-upgradeable] for more information.

Changes to the files in this directory should be commited as well.

Upgrading the contract

To upgrade the contract you should add a new migration file in the migrations/* directories increasing the migration number.

It looks like so:

require('dotenv').config({ path: "../.env" });

const PythUpgradable = artifacts.require("PythUpgradable");

const { upgradeProxy } = require("@openzeppelin/truffle-upgrades");

/**
 * Version <x.y.z>.
 * 
 * Briefly describe the changelog here.
 */
module.exports = async function (deployer) {
    const proxy = await PythUpgradable.deployed();
    await upgradeProxy(proxy.address, PythUpgradable, { deployer });
}

When changing the storage, you might need to disable the storage checks because Open Zeppelin is very conservative, and appending to the Pyth State struct is considered illegal. Pyth _state variable is a Pyth State struct that contains all Pyth variables inside it. It is the last variable in the contract and is safe to append fields inside it. However, Open Zeppelin only allows appending variables in the contract surface and does not allow appending in the nested structs.

To disable security checks, you can add unsafeSkipStorageCheck: true option in upgradeProxy call. If you do such a thing, make sure that your change to the contract won't cause any collision. For example:

  • Renaming a variable is fine.
  • Changing a variable type to another type with the same size is ok.
  • Appending to the contract variables is ok. If the last variable is a struct, it is also fine to append to that struct.
  • Appending to a mapping value is ok as the contract stores mapping values in a random (hashed) location.

Anything other than the operations above will probably cause a collision. Please refer to Open Zeppelin Upgradeable (documentations)[https://docs.openzeppelin.com/upgrades-plugins/1.x/writing-upgradeable] for more information.

Versioning

We use Semantic Versioning for our releases. When upgrading the contract, update the npm package version using npm version <new version number> --no-git-tag-version. Also, modify the hard-coded value in version() method in the Pyth.sol contract to the new version. Then, after your PR is merged in main, create a release like with tag pyth-evm-contract-v<x.y.z>. This will help developers to be able to track code changes easier.

Testing

The pyth-js repository contains an example with documentation and a code sample showing how to relay your own prices to a target Pyth network. Once you have relayed a price, you can verify the price feed has been updated by doing:

$ npx truffle console --network $MIGRATIONS_NETWORK
> let p = await PythUpgradable.deployed()
> p.queryPriceFeed("0xf9c0172ba10dfa4d19088d94f5bf61d3b54d5bd7483a322a982e1373ee8ea31b") // BTC Testnet or any other address

Verifying the contract

Please first try verifying the contract using truffle as described in VERIFY.md. It that doesn't work Try to manually verify the contract using the explorer UI. You can try to upload the standard json output in build/contracts directory. If that doesn't work either, you can flatten the contract and try to verify it.

To flatten the contract, run the following command:

npx sol-merger contracts/pyth/PythUpgradable.sol

It will create a new file PythUpgradable_merged.sol which you can use in the explorer to verify the implementation contract (using exact sol version and optimization flag). After verifying implementation, you can verify the proxy.

Troubleshooting

  • Sometimes the truffle might fail during the dry-run (e.g., in Ethereum). It is because openzeppelin does not have the required metadata for forking. To fix it please follow the suggestion here.

  • Sometimes due to rpc problems or insufficient gas the migration is not executed completely. It is better to avoid doing multiple transactions in one migration. However, if it happens, you can comment out the part that is already ran (you can double check in the explorer), and re-run the migration. You can avoid gas problems by choosing a much higher gas than what is showed on the network gas tracker. Also, you can find rpc nodes from here