[consensus] P90 latency test with faults + strict leader reputation#19325
Closed
danielxiangzl wants to merge 6 commits into
Closed
[consensus] P90 latency test with faults + strict leader reputation#19325danielxiangzl wants to merge 6 commits into
danielxiangzl wants to merge 6 commits into
Conversation
Replaces the land_blocking forge suite with a latency-focused test that uses a mainnet-representative validator distribution (~70% EU, ~20% NA, ~10% Asia) instead of the previous even 25%/25%/25%/25% four-region split. The even split over-weights Asia (25% vs ~2% on mainnet) and under-weights EU, making P90 thresholds misleading. The new distribution causes EU proposers to dominate rounds as they do on mainnet, exercising the actual latency bottlenecks (distant proposers racing with EU batch arrival). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Adds realistic_env_p90_latency_with_faults forge test that extends the mainnet-like P90 latency test with periodic validator failures. Kills the same 2 fixed validators for 15s, restarts them, waits 15s, then repeats throughout the test duration. Wires land_blocking to run this test so it triggers on PR CI. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This comment has been minimized.
This comment has been minimized.
danielxiangzl
force-pushed
the
daniel/latency-with-faults-strict-leader
branch
from
badfbea to
6921a13
Compare
When the test loop exits mid-cycle, validators may be stopped. The subsequent catchup check queries all nodes including stopped ones, causing the test to hang for 30+ minutes. Restart any stopped targets before returning. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add failure_window_num_validators_multiplier to ProposerAndVoterConfig, allowing the failure counting window to be configured independently from the proposer window. This prevents the oscillation problem where flaky validators cycle through failed→inactive→active states because failures age out of the short proposer window (~56s) before the validator comes back online. When set to 0 (default), falls back to proposer_window_num_validators_multiplier for backward compatibility. When set to a larger value (e.g., 50), failures are remembered for ~4.7 min, keeping chronically failing validators penalized even when they briefly come back online and vote. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…test The default value of 0 falls back to proposer_window (10x), making the strict leader reputation change a no-op. Set it to 50 so failed validators are remembered for ~4.7 min, actually exercising the feature. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
danielxiangzl
force-pushed
the
daniel/latency-with-faults-strict-leader
branch
from
6921a13 to
5b26e3a
Compare
Contributor
Forge is running suite
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
daniel/latency-with-faults(P90 latency forge test with periodic validator failures) withdaniel/strict-leader(separate failure window for leader reputation)Test plan
land_blocking(wired torealistic_env_p90_latency_with_faults) and verify latency thresholds passdaniel/latency-with-faultsbaseline (without strict leader)🤖 Generated with Claude Code