How to Configure a BuildKit Dashboard for Monitoring Image Build Latency, Success Rate, and Error Rate?

[CaptD6]

Hello everyone,

I've been using BuildKit to build container images and I want to set up a monitoring dashboard for our operations team. The goal is to track key performance indicators (KPIs) such as build latency, success rate, and error rate more effectively.

I have a few specific questions and would appreciate guidance from the community:

Build Latency: How can I obtain data on the latency of each image build? Are there recommended tools or methods to capture and visualize this data?

Success and Error Rates: Does BuildKit support automatically logging the number of successful and failed builds? Are there existing solutions that can generate reports on these statistics?

Logs and Data Format: To integrate these metrics into our existing monitoring system (such as Prometheus or Grafana), what logs or data formats should we focus on?

Additional Suggestions: Are there other metrics worth monitoring, or any configurations or tools you recommend for enhancing operational efficiency?

Thank you very much for your help! I'm sure other users with similar needs will also benefit from this information.

I look forward to your suggestions and shared experiences.

Thanks!

[CaptD6]

By the way, I deployed buildkitd stand-alone on DaemonSet. And I deploy with following config:

env:
  OTEL_SERVICE_NAME: "buildkit-test"
  OTEL_TRACES_EXPORTER: "otlp"
  OTEL_EXPORTER_OTLP_ENDPOINT: "otlp://xxx"
  OTEL_EXPORTER_OTLP_TRACES_ENDPOINT: "otlp://xxx"
  OTEL_EXPORTER_OTLP_INSECURE: "true"
  OTEL_METRICS_EXPORTER: "otlp"

[profnandaa]

See if this can give some ideas on where to start - https://docs.docker.com/build/debug/opentelemetry/

[CaptD6]

Thanks for your reply! I have successfully collected the trace data, but I would like to monitor more metrics. These metrics could be used to configure alerts and are very helpful for operations work.

[CaptD6]

@tonistiigi Can you help plz?

[tonistiigi]

BuildKit supports History APIs where you can access all these properties for old builds that you can then collect into your statistics tool.

Docker Desktop Build UI (that pulls data via History API) can also show you build stats over time https://docs.docker.com/desktop/use-desktop/builds/#build-history .

[CaptD6]

That's great! Could you please point me to the official documentation for the BuildKit History APIs? I'd like to understand how to programmatically access historical build metadata and integrate it with our monitoring system. :)