# Self-Healing CI: Recovering When the Runner Briefly Loses Communication

> A job that fails because the runner briefly lost contact with the control server is a network blip, not a code bug. See the manual fix and how self-healing CI reruns it.

Source: https://latchkey.dev/learn/self-healing-ci/self-healing-runner-lost-communication  
Updated: 2026-06-26

A job marked failed because the runner briefly lost contact with the control server is an infrastructure blip, not your build failing -- a rerun completes it.

## What makes a failure safely retryable

Automatic retry is only correct for failures that are genuinely transient. Retrying a deterministic failure wastes minutes and hides a real defect, so the classification matters more than the retry mechanism.

- Safe to retry: network timeouts, registry 5xx, transient DNS failures, a service container that was not ready, a spot instance reclaimed mid-run.
- Not safe to retry: assertion failures, compile errors, lint violations, anything that fails identically on every attempt.
- Ambiguous, and worth investigating rather than retrying: out-of-memory kills, disk exhaustion, and flaky tests. These repeat under load and a retry only hides the trend.
- Always record that a retry happened. A pipeline that silently retries is a pipeline whose real failure rate you do not know.

## FAQ

### What causes Self-Healing CI: recovering when the runner briefly loses communication?

A job is failed because the runner momentarily lost its connection to the CI control server -- a brief network partition or heartbeat miss -- even though the work itself was healthy. The code is fine; the runner just dropped contact for a moment. A human re-runs the job and it completes unchanged.

### How do I fix Self-Healing CI: recovering when the runner briefly loses communication manually?

[object Object]

### Can Self-Healing CI: recovering when the runner briefly loses communication be fixed automatically?

A lost-communication failure has an unmistakable infrastructure signature -- a connectivity drop, not an application error -- and the safe response is to rerun on a healthy runner. A self-healing CI pipeline recognizes the communication loss, reruns the job, and only counts a failure that reproduces with healthy connectivity, so a

---

Latchkey runs CI/CD that repairs its own failures. Agent entry points: https://latchkey.dev/agent.txt, https://latchkey.dev/openapi.json, https://latchkey.dev/llms.txt