All validators broadcast commit vote messages

[vusirikala]

Description

Currently, all the validators send their commit votes to the block proposer. The block proposer aggregates the commit votes and broadcasts the commit decision. In order to reduce one hop of latency, this PR will let all the validators directly broadcast their commit votes.

Test Plan

The goal of the PR is to reduce latency. We ran forge test with 100 nodes on this PR vs main branch.

Forge run on this PR: https://github.com/aptos-labs/aptos-core/actions/runs/7981466686
Forge run on main: https://github.com/aptos-labs/aptos-core/actions/runs/7965951082

[trunk-io]

⏱️ 51h 18m total CI duration on this PR

Job	Cumulative Duration	Recent Runs
rust-unit-tests	10h 20m	🟥 ⬜ 🟩 ⬜ 🟩 (+12 more)
windows-build	5h 24m	🟩 🟩 🟩 🟩 🟩 (+12 more)
rust-unit-coverage	5h 9m	⬜ ⬜ 🟩
execution-performance / single-node-performance	4h 56m	🟩 🟥 🟩 🟩 🟩 (+10 more)
rust-smoke-tests	4h 53m	🟩 ⬜ 🟩 ⬜ 🟩 (+9 more)
rust-smoke-coverage	4h 14m	⬜ ⬜ 🟩
forge-e2e-test / forge	2h 41m	🟥 🟥 🟩 🟩 🟩 (+9 more)
rust-images / rust-all	2h 27m	🟩 ⬜ 🟩 ⬜ 🟩 (+11 more)
forge-compat-test / forge	2h 5m	🟩 🟩 🟩 🟩 🟩 (+6 more)
cli-e2e-tests / run-cli-tests	1h 47m	🟥 🟥 🟥 🟥 🟥 (+7 more)
run-tests-main-branch	1h 36m	🟥 🟥 🟥 ⬜ 🟥 (+12 more)
rust-lints	1h 24m	🟥 ⬜ 🟩 ⬜ 🟩 (+12 more)
adhoc-forge-test / forge	1h 21m	🟩 🟩
check	1h 2m	🟩 ⬜ 🟩 ⬜ 🟩 (+12 more)
general-lints	35m	🟩 🟩 🟩 ⬜ 🟩 (+12 more)
check-dynamic-deps	34m	🟩 🟩 🟩 🟩 🟩 (+12 more)
indexer-grpc-e2e-tests / test-indexer-grpc-docker-compose	16m	🟩 🟥 🟩 🟩 🟥 (+5 more)
node-api-compatibility-tests / node-api-compatibility-tests	9m	🟩 🟩 🟩 🟩 🟩 (+5 more)
semgrep/ci	7m	🟩 🟩 🟩 🟩 🟩 (+12 more)
file_change_determinator	3m	🟩 🟩 🟩 🟩 🟩 (+12 more)
file_change_determinator	3m	🟩 🟩 🟩 🟩 🟩 (+12 more)
file_change_determinator	3m	🟩 🟩 🟩 🟩 🟩 (+11 more)
execution-performance / file_change_determinator	2m	🟩 🟩 🟩 🟩 🟩 (+9 more)
permission-check	1m	🟩 🟩 🟩 🟩 🟩 (+12 more)
permission-check	52s	🟩 🟩 🟩 🟩 🟩 (+12 more)
determine-docker-build-metadata	49s	🟩 🟩 🟩 🟩 🟩 (+11 more)
permission-check	49s	🟩 🟩 🟩 🟩 🟩 (+12 more)
permission-check	43s	🟩 🟩 🟩 🟩 🟩 (+11 more)
permission-check	41s	🟩 🟩 🟩 🟩 🟩 (+12 more)
determine-forge-run-metadata	12s	🟩 🟩
upload-to-codecov	12s	🟩

_{settings ⋅ feedback ⋅ docs ⋅ learn more about trunk.io}

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (ea133dc) 71.4% compared to head (60744f1) 69.9%.
Report is 21 commits behind head on main.

❗ Current head 60744f1 differs from pull request most recent head cb38636. Consider uploading reports for the commit cb38636 to get more accurate results

Additional details and impacted files

@@             Coverage Diff             @@
##             main   #12059       +/-   ##
===========================================
- Coverage    71.4%    69.9%     -1.5%     
===========================================
  Files         802     2222     +1420     
  Lines      184373   419549   +235176     
===========================================
+ Hits       131737   293517   +161780     
- Misses      52636   126032    +73396     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ibalajiarun]

You need to adjust the timeout here to make the test work for 100 nodes.

aptos-core/testsuite/forge/src/backend/k8s/swarm.rs

Line 709 in cca7554

let timeout_duration = Duration::from_secs(30);

Bump it to 5 minutes or so in your PR.

[sitalkedia]

Can you add a TODO to remove the proposal commit decision broadcast in following release?

[github-actions]

✅ Forge suite compat success on aptos-node-v1.9.5 ==> cb386360e848aba2bc230f60d10718e9c49f77df

Compatibility test results for aptos-node-v1.9.5 ==> cb386360e848aba2bc230f60d10718e9c49f77df (PR)
1. Check liveness of validators at old version: aptos-node-v1.9.5
compatibility::simple-validator-upgrade::liveness-check : committed: 6761 txn/s, latency: 4943 ms, (p50: 4800 ms, p90: 8100 ms, p99: 9600 ms), latency samples: 236660
2. Upgrading first Validator to new version: cb386360e848aba2bc230f60d10718e9c49f77df
compatibility::simple-validator-upgrade::single-validator-upgrade : committed: 1669 txn/s, latency: 16700 ms, (p50: 17700 ms, p90: 21700 ms, p99: 22200 ms), latency samples: 85160
3. Upgrading rest of first batch to new version: cb386360e848aba2bc230f60d10718e9c49f77df
compatibility::simple-validator-upgrade::half-validator-upgrade : committed: 109 txn/s, submitted: 327 txn/s, expired: 218 txn/s, latency: 24339 ms, (p50: 7900 ms, p90: 90200 ms, p99: 90300 ms), latency samples: 16034
4. upgrading second batch to new version: cb386360e848aba2bc230f60d10718e9c49f77df
compatibility::simple-validator-upgrade::rest-validator-upgrade : committed: 2609 txn/s, latency: 11366 ms, (p50: 12100 ms, p90: 17500 ms, p99: 17800 ms), latency samples: 117420
5. check swarm health
Compatibility test for aptos-node-v1.9.5 ==> cb386360e848aba2bc230f60d10718e9c49f77df passed
Test Ok

• Grafana dashboard
• Humio Logs
• Axiom Logs
• Test runner output
• Test run is land-blocking

[github-actions]

✅ Forge suite realistic_env_max_load success on cb386360e848aba2bc230f60d10718e9c49f77df

two traffics test: inner traffic : committed: 7628 txn/s, latency: 4997 ms, (p50: 4500 ms, p90: 6200 ms, p99: 11400 ms), latency samples: 3295620
two traffics test : committed: 100 txn/s, latency: 2198 ms, (p50: 2000 ms, p90: 2300 ms, p99: 7000 ms), latency samples: 1760
Latency breakdown for phase 0: ["QsBatchToPos: max: 0.246, avg: 0.206", "QsPosToProposal: max: 0.184, avg: 0.155", "ConsensusProposalToOrdered: max: 0.548, avg: 0.533", "ConsensusOrderedToCommit: max: 0.395, avg: 0.376", "ConsensusProposalToCommit: max: 0.942, avg: 0.909"]
Max round gap was 1 [limit 4] at version 873472. Max no progress secs was 4.821868 [limit 15] at version 3261394.
Test Ok

• Grafana dashboard
• Humio Logs
• Axiom Logs
• Test runner output
• Test run is land-blocking