synth-bench: update nonces on disk independently of tx outcome

[mooori]

Context

synth-bm was initially built for high throughput workloads of simple transactions like native token transfers with wait_until: NONE. In such a setting, all RPC calls are expected to succeed with successful outcomes. Hence, to avoid slowing down TPS, account nonces are updated on disk only after all transactions succeeded.

Problem

When benchmarking more complex function calls with wait_until: EXECUTED_OPTIMISTIC, transactions with error outcome might occure as part of regular workload execution flow. So account nonces should be updated independently of transaction outcomes. Otherwise nonces on disk become outdated and benchmark runs re-using the same account will fail with nonce errors.

Implementation

This is how nonces are updated currently.

Nonce updates on disk should happen in the background to avoid reducing the number of transactions the tooling can generate. This approach could take inspiration from RPCResponseHandler which runs in a tokio task and receives work via a channel.

[nadongjun]

@mooori I’d like to take this on—please assign it to me if possible. Thanks!

[mooori]

@nadongjun thank you for your interest in contributing! Though we are currently considering to work on this item ourselves. So please hold off for now, to avoid duplicate work.