Add tool to test application of txs to ledger

[georgeee]

Benchmark-like tool that creates a set number of 9-account-updates transactions and applies them to a given ledger.

Explain how you tested your changes:

• Ran a tool against mainnet-like ledger and measure performance

How to use the tool:

MINA_PRIVKEY_PASS= mina ledger test-apply --ledger-path ledger --privkey-path key --num-txs 128

Checklist:

• Dependency versions are unchanged
  • Notify Velocity team if dependencies must change in CI
• Modified the current draft of release notes with details on what is completed or incomplete within this project
• Document code purpose, how to use it
  • Mention expected invariants, implicit constraints
• Tests were added for the new behavior
  • Document test purpose, significance of failures
  • Test names should reflect their purpose
• All tests pass (CI will check this if you didn't)
• Serialized types are in stable-versioned modules
• Does this close issues? None

[emberian]

Maybe use the logger above for all these eprintf? what do you think?

[emberian]

Looks good. I don't know where to get a serialized block myself. Should we have this run in CI (with small parameters)?

[georgeee]

@emberian I was using block serializations we get from the uptime system, but probably there are less exotic ways to get it.

I definitely want to come up with a more friendly flow when start working on #14800

[georgeee]

How to run the benchmark

Assuming mina is built with devnet profile:

MINA_PRIVKEY_PASS= mina ledger test-apply --num-txs 128 --ledger-path ledger --privkey-path key

where:

• ledger is an unpacked directory with contents of a genesis ledger from the archive.
• key is a secret key from file

The output will look like:

Init root jxDivUg42TF8bYRCojgBAcXZkCRWpBVLoQcBy8u3kdi9uqNgaCu                                                                                                                               Result of application 0: true (took 42.978286743164062ms): new root jxx9niKXgim3fijPFh7GeHpvbJUGtoorC7aaXLtbBThuS7weoEE
Result of application 1: true (took 53.806781768798828ms): new root jxzUn1JXpRptqZ2zct2Tfgr5hF9xsQL38cxYWs2f4AT5qayUKPg
Result of application 2: true (took 48.964262008666992ms): new root jxeQbrb68q4efgivqZz5XX6vuzC68uyAuyJqqWcJdTUSKjwcnJ8
Result of application 3: true (took 52.176713943481445ms): new root jxncmbctCCZRpCZvRLrGj8rk174r4MR3YSS5ZqRaLvjwnrWXAhw
Result of application 4: true (took 43.39289665222168ms): new root jx5NpQz3f8sF63yjidmXE8QTv91jF9SRPHY47Zb59NNDDtHxUXH
....
....
....
Result of application 574: true (took 60.74070930480957ms): new root jwdgzX3ufDjra967AURYai1W9Gk8MKb7Pu4LmtkrZaqc9Fv9J7e
Result of application 575: true (took 60.318946838378906ms): new root jx3HRNFQTMTRWp9o8YNBrfqWcAyH5LT8BcuQT74a77M1QXWHVp7
Result of application 576: true (took 59.606313705444336ms): new root jwojN8v5nLyYsPkiNsrF8g8gqHKiwKpiN5JEikZBFLk1dgv9DmN
Result of application 577: true (took 56.887388229370117ms): new root jxbioQSEwPesBpQN8tNrV2CTXcDDvnXcJ3rWfuX7C9ZaEmRPa6V
Result of application 578: true (took 55.930376052856445ms): new root jxFTQKgTa3HrY6UBvmmY5bpwYSUNAsQmR5dcQrJ2VunLfg6gA3m
Result of application 579: true (took 62.193393707275391ms): new root jwbw8BWqnM41BBsvnyXeeghGCPzcgjhAXBtejeDfC3tKG9b5sKi
Result of application 580: true (took 2.4975223541259766s): new root jx6HwhECxAme3ycDu7ZyyEMfoV63eHE4Y1ahzbQfDqHpxRrhHbc

Benchmark is structured in two phases:

• Preparatory phase: apply 580 (controlled via --rounds) transaction payloads to the ledger:
  • each in the same way as if these represented transactions of a block
  • each payload consisting of 3 transactions (controlled via --num-txs-per-round)
• Measurement phase: apply a single payload of N transactions (with N controlled via --num-txs)
  • each transaction consisting of 8 signature-backed account updates (+ one more account update for the fee payer)

So the benchmark simulates a single pass of ledger application of a full block. This is not a real benchmark, ina way that we run this only once. In future iterations I want to change the format and use the core_bench, but I think this PR is good to be merged as is.

[georgeee]

!ci-build-me