[General] Sepolia transactions occasionally failing

[MichaelSheinmanCircle]

Environment

Testnet

Provide a brief description of the functionality you're trying to implement and the issue you are running into.

Seeing an issue with sepolia transaction occasionally failing with an error "Error function_selector = 0x, data = 0x". This happens approximately once for every 50 transactions I send to the network.

Additional Details

Example transactions:
https://sepolia.explorer.zksync.io/tx/0x7183effad349b128ef82d58eb27e6b5cdcd2d4f02ae37c8cf582a83136db1345
https://sepolia.explorer.zksync.io/tx/0xf9fb1d51b4492b288eade87f310a15c912439e3d7d52e3cc67a3953a98f2a5d6

I heard these issues tend to indicate gas problem, however I am using the eth_gasPrice and the eth_estimateGas RPC calls for the gas estimation, and also adding a buffer on these values for extra caution.

Expected behaviour would be either that the transaction fails with some error message when I send the transaction, or that the transaction succeeds. However, the current behaviour is that the sending part of the transaction goes through and then the transaction fails.

[bxpana]

Thank you for moving that 😁 Will have the team take a look and see if we can find something or hopefully someone else from the community having a similar issue can weigh in.

[bxpana]

@MichaelSheinman how are things going here? Have you run into these same issues on testnet lately?

[MichaelSheinmanCircle]

Yes, here is a recent example from 5 days ago: https://sepolia.explorer.zksync.io/tx/0x0d97de2763781cf36a7052d7655ef292327c59dccc84fe3747a89660b140b821

[MichaelSheinmanCircle]

Note that this case is a very odd race condition - our system will pick up the transaction from the blockchain successfully, and then for some reason the transaction will fail later before it's officially completed in ethereum.
Perhaps a good start here is to update the error message shown on the explorer so we get some more descriptive information here.

[MichaelSheinmanCircle]

FYI got more failures than usual earlier today, not sure if it's related to an update
https://sepolia.explorer.zksync.io/tx/0xbe205d1885163823ccf7ed939313808fbd01c9a05efeb93b2473f3fe893aee35
https://sepolia.explorer.zksync.io/tx/0xfae8be164bcf0b284338a5326bf0e3c9454083f2ddcd37454e9feafb78e1dce4
https://sepolia.explorer.zksync.io/tx/0xa8acc6eb7166b204f2ce514eae28326dfbcfae481fb78f5978fbd4addcb7d854

[hbriese]

@bxpana any updates? Still seeing this quite frequently

[MichaelSheinmanCircle]

same

[bxpana]

@hbriese and @MichaelSheinman nothing yet. We're seeing some volatility in gas on Sepolia that we're trying to understand better. Right now the best suggestion is to increase the gas limit. We're looking into improving the era-test-node to help with understanding these txns.

[bxpana]

Circling back, this is usually caused by the underlying L1 gas spike, which causes the pubdata price to skyrocket. This then leads to the txn failing. Things seem to be more stable now, but if you start to see any issues again let us know and we can reopen this thread.