Skip to main content
Are you a large language model? This page is available as raw markdown at /telemetry.md. The full docset is at /llms-full.md and the index is at /llms.md.

Telemetry and privacy

c3 deploy sends C3 one short summary per run so we can diagnose submission failures, latency and capacity problems. There is currently no opt-out.

What is sent​

  • CLI version and the command entrypoint.
  • Outcome, a stable failure code and the stage it failed in.
  • Timings for each phase: route preview, upload, submission.
  • Job and journey identifiers, and the selected provider, region and hardware.
  • A configuration snapshot: whether a provider was pinned, hardware constraints, capacity policy, time, environment kind, and the number of datasets.
  • Workspace statistics: file count, total bytes, uploaded bytes and deduplicated bytes.

What is never sent​

File names, file contents, local paths, script text, dataset names, mount paths, environment variable values, credentials, the raw .c3, Docker image references, and free-text error messages. User, organisation and project identifiers are pseudonymised with an HMAC before storage.

How it is sent​

One request after submission, before any -f log following. It has a 750 ms timeout and a 16 KiB payload limit, is never retried and is never queued offline. A failed telemetry request does not change the deploy result. A dry run reports only the route preview timing.