feat: measuring node round trips

[benesjan]

In this PR I implement round trip tracking and I use this info as a basis for a roundtrip regression test. This check was placed in yarn-project/end-to-end/src/bench/client_flows/amm.test.ts. This test was a good candidate for this because we already use it for benchmarking and it doesn't communicate with node over http (instead we have a direct access to the instance) - this is great as it makes the dest more deterministic and hence the roundtrip counts should not be brittle (when communicating over network I can imagine that latency or something could result in different counts e.g. when syncing PXE or something - not super sure if this is still the case though but it was the case in the past when we were polling node for new blocks)

In that test I also implemented a debugging functionality that allows for debugging the regression by nicely printing out round trip trace.

To track the round trips I updated ProxiedNode by having an inFlightCount counter tracking num current RPC calls. This allows us to group parallel rpc calls into the same round trip.

Fixes https://linear.app/aztec-labs/issue/F-230/improve-benchmarks

The linked issue seems to have a wider scope than just the round trip measurements so not sure if we want to fully close it with this PR.

[benesjan]

• feat: measuring node round trips #19291 👈 (View in Graphite)
• next

This stack of pull requests is managed by Graphite. Learn more about stacking.

[mverzilli]

Approving as I don't see anything that could break, and I ultimately trust your judgment on whether this adds value as-is.

IMO there's problems in how we're measuring things here, and the names we're using to collect those stats can hide those problems. But maybe we find some big issues using this, fix them, archive the whole topic so 🤷‍♂️

[AztecBot]

Flakey Tests

🤖 says: This CI run detected 4 tests that failed, but were tolerated due to a .test_patterns.yml entry.

\033FLAKED\033 (8;;http://ci.aztec-labs.com/806774aa44d4fc6a�806774aa44d4fc6a8;;�):  yarn-project/end-to-end/scripts/run_test.sh simple src/e2e_p2p/gossip_network_no_cheat.test.ts (51s) (code: 1) group:e2e-p2p-epoch-flakes (\033Jan Beneš\033: feat: measuring node round trips (#19291))
\033FLAKED\033 (8;;http://ci.aztec-labs.com/165bc6912882e63c�165bc6912882e63c8;;�):  yarn-project/end-to-end/scripts/run_test.sh simple src/e2e_epochs/epochs_invalidate_block.parallel.test.ts "committee member invalidates a block if proposer does not come through" (96s) (code: 1) group:e2e-p2p-epoch-flakes (\033Jan Beneš\033: feat: measuring node round trips (#19291))
\033FLAKED\033 (8;;http://ci.aztec-labs.com/21240767ef2fbde6�21240767ef2fbde68;;�):  yarn-project/end-to-end/scripts/run_test.sh simple src/e2e_epochs/epochs_invalidate_block.parallel.test.ts "proposer invalidates multiple blocks" (131s) (code: 1) group:e2e-p2p-epoch-flakes (\033Jan Beneš\033: feat: measuring node round trips (#19291))
\033FLAKED\033 (8;;http://ci.aztec-labs.com/10affe6cd855a8c1�10affe6cd855a8c18;;�):  yarn-project/end-to-end/scripts/run_test.sh simple src/e2e_p2p/multiple_validators_sentinel.parallel.test.ts "collects attestations for validators in proposer node when block is not published" (112s) (code: 1) group:e2e-p2p-epoch-flakes (\033Jan Beneš\033: feat: measuring node round trips (#19291))

[mverzilli]

Cool! Nothing blocking, the only one I would pay some love to is the comment about not needing to use http to communicate with node in this context, which surprised me

[benesjan]

The e2e_bot flake has already been discussed here so I am re-adding this to queue.