Spans are already flowing into Tempo from the OpenFeature TracesHook , but the metrics half is dead — the MeterProvider has no exporter and the MetricsHook was never registered. The dashboard the operator wants to triage from is empty. The k6 loadgen is idle, waiting for a flag flip to turn it on.
A huge thank you to @simon.schrottner for creating this challenge!
What You’ll Learn
- How the OpenFeature OpenTelemetry hooks (
TracesHookandMetricsHook) join flag evaluations to the rest of an application’s telemetry without a separate ingestion path - How to author your own
Hook— a tiny class that copies merged-eval-context attributes onto the active OTel span — to close the loop between why a flag resolved the way it did and what the operator sees in Tempo - How
fractionalrollout in flagd buckets users bytargetingKey— same key, same bucket, every request — and how to read that bucketing off a dashboard - How a flag flip is a faster operational lever than a redeploy when a rollout is misbehaving — the difference between a one-line config change and a twenty-minute deployment
Rewards
A 50% Linux Foundation certification voucher for 1st place and Credly badges for the top 3 — for players who complete all three levels by Tuesday, 26 May 2026 at 23:59 CET. See the adventure overview for ranking rules and eligibility details.
Get Started
Discussion
Share your solutions, your questions, and thoughts. We look forward to read from you.
Other levels
Beginner: Stand up the lab
Intermediate: Outcome by cohort
