Going Plaid: Striving for Speed of Light in CI pipelines
summary
Adds telemetry to GitHub Actions CI, exporting GitHub run metadata as traces to Tempo + Grafana, using OpenTelemetry SDKs to parse job and step timings into spans. Quantifies startup wait, container image setup, cache hits, and matrix synchronization bottlenecks and surfaces GPU resource hogs. Builds dashboards and a dispatch actions pattern that checks out shared action code and preserves env/secrets. Measures drive targeted fixes and design tradeoffs though definitive cost savings are not yet claimed.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 0:57Speed of light in CI
- 3:02CI optimization purpose
- 7:19CI observability pillars
- 9:30Instrumenting builds
- 12:02CI traces in Grafana
- 13:49Span traits and bottlenecks
- 15:55CI performance tracking
- 16:58Reusable workflows
- 19:30GitHub Actions limits
- 22:02Secrets in shared workflows
- 24:52Dispatch actions
- 26:21Environment variables in CI
- 28:49Conclusion & Q&A
related talks
30mSupercharged GitHub Actions monitoring with OpenTelemetry at Elastic
44mSupercharged GitHub Actions monitoring with OpenTelemetry at Elastic
28mEnd-to-end CI/CD observability with open source
29mFrom Flaky Builds to Self-Healing Deployments: An OTel-Driven CI/CD...
43mCI/CD with Github Actions
13mLightning Talk: Observability-First DevSecOps: Building Resilient Multi-Cloud P...
24mUnlocking CI Performance in the Age of Agentic Development
11mTracing CI/CD with OpenTelemetry