Closed glevco closed 1 day ago
Branch | refactor/bench-data |
Testbed | ubuntu-22.04 |
Benchmark | Latency | Benchmark Result nanoseconds (ns) (Result Δ%) | Lower Boundary nanoseconds (ns) (Limit %) | Upper Boundary nanoseconds (ns) (Limit %) |
---|---|---|---|---|
sync-v2 (up to 20000 blocks) | 📈 view plot 🚷 view threshold | 101,471,694,713.76 (-0.55%) | 91,826,613,182.35 (90.49%) | 112,232,527,222.88 (90.41%) |
All modified and coverable lines are covered by tests :white_check_mark:
Project coverage is 84.36%. Comparing base (
7f53032
) to head (63d11c1
). Report is 1 commits behind head on master.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
@jansegre
Should we call it something else other than "sync_v2" now? Using load-from-logs shouldn't trigger any sync code, right? Maybe "load_sample". Other than that it looks good to me.
The benchmark is still the same, it uses Sync-v2. I only changed the way the server node boots up (from syncing with the network vs from logs). Then, a second node connects to the server node and the measurement is made on the sync of that second node.
Motivation
Currently, CI benchmark directly connects to testnet to download benchmark data (it performs a sync all the way up to 20k blocks). This PR changes it to use the latest load-from-logs CLI tool update, making the CI run faster and more reliably. Then, we'll also be able to introduce new CI benchmarks in other PRs, including for example syncing a "dummy" network with more transactions than blocks.
Acceptance Criteria
Checklist
master
, confirm this code is production-ready and can be included in future releases as soon as it gets merged