The Dashboard That Averaged Averages

The latency panel is a flat green line. Synthetic probes pass. Real users hit 1s tails. Someone 'optimized' the percentile calculator by averaging batch means so the chart looked smoother in the exec review.

Initial failing signalp95 hides a 1000ms tail spikeExpected behavior: The p95 latency reflects the 1000 ms tail spike. · actual behavior: The p95 latency is about 10 ms and hides the tail spike.

Progress

1Reproduce the bugRun checks

2Edit the fixOpen

3Pass checksOpen

4Submit fixOpen

Reproduction Steps

Build a sample list with a rare high tail.
Compute p95.
Confirm the result reflects the tail, not the batch means.

Focus Mode

Start

src/dashboard.py
src/latency.py

Likely area

src/latency.py
src/dashboard.py
src/metrics_store.py

Challenge Coach

Hints reveal broad direction first, then narrow toward the suspect path without showing hidden tests or exact solution code.

TopicObservability

DifficultyMedium

Est. time15 min

Workspacepreparing

Explorer

Loading files

Evidence Map

Failing signal loaded from challenge API
Workspace files loaded from workspace API
Results rendered only after API run/submit

Autosaved

workspace

Outputread-onlyready$ prepare workspace

Preparing workspace