DanielVF
commented
2 years ago Just more data on this:
My deploy failures have been very similar:
- During large deploy scripts
- After deploying contracts, near the start of the part where we send deployer transactions to those just deployed contracts (The middle phase of our typical deploy script)
- On mainnet (as opposed to fork)
In the the two failure I've had the script process just hang, with no further messages. I don't remember if it just didn't start the next transaction, or hung during a transaction.
I have opted to manually patch the deploy script when this has happened, since because of the size of the deploys these happened in and gas prices at the time, we'd already spent an eth or three on deploying the contracts.
Occasionally when we run the hardhat script for deploying new contracts on Mainnet, we run into errors. Root cause varies but can in my experience be attributed to
In case of such a failure, the operator typically has to manually figure at what step the script failed (using console logs or by checking blockchain txs), comment out the portion of the script already executed, modify the script (for ex. to read addresses of contracts already deployed in the prior failed run) and re-run the script. This can get a bit painful.
It would be nice if we could enhance our current deploy script framework to have some sort of intermediate "checkpoints". And allow to restart where the script failed.
Note: this may complicate things quite a bit in terms of the deploy logic. It may not be worth the extra complexity and engineering time. The alternative is to not worry about partial recovery and just pay the extra gas costs and re-run the entire script... :)