nonce and thread-safety

[mattador1]

Do not know if this fully fits ethers discussion, but as I am using it all the time, maybe somebody will know.
If you are using in node.js for example 3 worker threads and and every minute you execute 3 jobs which initiate transaction all at the same time at same time to for example hardhat node (localhost). I see that I got NONCE_EXPIRED on some of transactions. Using ethers v6. Any suggestion how to handle this situation, what would be best way to ensure that nonce isn't repeated between worker threads.

[ricmoo]

I haven’t use workers in node.js much, but the best way is likely to keep the current nonce in the main thread, and use postMessage to communicate back and forth with the main thread, which will increment the value and return the old value each time a child worker asks for it.

I’m not sure if you can share an instance of a NonceManager within workers, but if so, that would be a much simpler solution. :)

[mattador1]

I tried first approach seems working clean, but do you know what if tx fails, than nonce is incremented, do I just leave it and move forward, this could introduce nonce gaps, is this problematic from your experience?

I tried setting setIntervals and just try and try again, eventually it takes correct nonce, it works but to me it seems messy. In general it seems when testing with hardhat that when I submit from multiple threads at the same time transactions that it somehow mess the nonce thingy as if hardhat node doesn't register fast enough to provide info to ethers about next nonce

[mattador1]

Also wondering, how ethers handles nonce increments, afaik, nonce is incremented when tx is mined. Does ethers have internal mechanism to increment nonce as default when I submit 2 tx one after another (without waiting to be mined), are those tx already generated with different nonce?