Dagger "failed to start engine" in CI

By Daniel Zoghalchali·Latchkey

The Dagger CLI created the engine container but it exited or never became ready. This is distinct from a connection failure: Docker is present, but the engine itself could not initialize because of privileges, resources, or a bad state.

What this error means

dagger fails with "failed to start engine" and the engine container exits shortly after creation, or never reports ready.

Error: failed to start engine: container dagger-engine exited: buildkitd: failed to create runc rootless: operation not permitted

Common causes

The engine lacks required privileges

The Dagger Engine embeds BuildKit and needs privileges to set up its runtime. An unprivileged container blocks initialization.

A corrupt engine state or resource limit

A stale engine volume or an out-of-resource runner can keep the engine from starting cleanly.

How to fix it

Run the engine privileged

1. Ensure the engine container runs privileged.
2. Remove a stale engine container and volume to reset state.
3. Re-run so the CLI provisions a fresh engine.

docker rm -f dagger-engine
docker volume rm dagger-engine-state || true
dagger call build

Inspect the engine logs

Read the engine container logs to see the exact initialization failure.

docker logs dagger-engine

How to prevent it

• Grant the engine the privileges it needs.
• Reset a stale engine volume when the engine wedges.
• Size runners for the engine plus your pipeline.

Frequently asked questions

Related guides

References

• Docker documentation
• Docker build cache
• GitHub Actions documentation