dagger/dagger · error

failed to start container: %w

Error message

failed to start container: %w

What it means

This wrapped error is produced by the image driver's create method when a leftover engine container matching the target name exists but ContainerStart fails to start it. The %w wrapping preserves the underlying backend error (docker/podman start failure) for errors.Is/As inspection. It means the driver found an existing engine and tried to reuse it, but the runtime refused.

Source

Thrown at engine/client/drivers/container.go:267

		}
		// run the container using that id in the name
		containerName = containerNamePrefix + id
	}

	leftoverEngines, err := d.collectLeftoverEngines(ctx, containerName)
	if err != nil {
		if errors.Is(err, context.Canceled) {
			return nil, err
		}
		slog.Warn("failed to list containers", "error", err)
		leftoverEngines = []string{}
	}

	for i, leftoverEngine := range leftoverEngines {
		// if we already have a container with that name, attempt to start it
		if leftoverEngine == containerName {
			if err := d.backend.ContainerStart(ctx, leftoverEngine); err != nil {
				return nil, fmt.Errorf("failed to start container: %w", err)
			}
			d.garbageCollectEngines(ctx, opts.cleanup, nil, slices.Delete(leftoverEngines, i, i+1))
			return &url.URL{Host: containerName}, nil
		}
	}

	// ensure the image is pulled
	exists, err := d.backend.ImageExists(ctx, opts.imageRef)
	if err != nil {
		return nil, fmt.Errorf("failed to inspect image: %w", err)
	}
	if !exists {
		if err := d.backend.ImagePull(ctx, opts.imageRef); err != nil {
			return nil, fmt.Errorf("failed to pull image: %w", err)
		}
	}

	volume := distconsts.EngineDefaultStateDir

View on GitHub (pinned to 82ba2681db)

Solutions

1. Remove the stale container (`docker rm -f <container>`) and let Provision create a fresh one
2. Check `docker start <container>` manually to see the raw runtime error
3. Restart the container daemon (Docker Desktop / podman service) if it is in a bad state
4. Check disk space and image availability; `docker system prune` if disk pressure caused start failure

Example fix

// before (stale container won't start)
docker start dagger-engine-abcdef  # fails: image not found
// after
docker rm -f dagger-engine-abcdef  # then re-run dagger, which provisions a new engine

engines, err := backend.ContainerLs(ctx)
if err == nil && slices.Contains(engines, containerName) {
    if err := backend.ContainerStart(ctx, containerName); err != nil {
        backend.ContainerRemove(ctx, containerName) // allow fresh create
    }
}

if err := backend.ContainerStart(ctx, name); err != nil {
    _ = backend.ContainerRemove(ctx, name) // clear stale container
    return nil, fmt.Errorf("failed to start container: %w", err)
}

When it happens

Trigger: create finds leftoverEngines containing exactly containerName and calls d.backend.ContainerStart(ctx, leftoverEngine), which returns a non-nil error (e.g. container is in a broken/removing state, runtime daemon unavailable, image missing for a stopped container).

Common situations: A stale engine container from a previous version whose image was pruned; container stuck in a restarting/exited state after a crash; Docker daemon partially responsive (exists in ls but cannot start).

Related errors

• failed to run container: %w
• container already exists
• install dnsmasq: %w
• install cni: %w
• mount does not exist

AI-assisted analysis of dagger/dagger@82ba2681db (2026-09-05). Data as JSON: /api/errors/d508c6253b288ee0. Report an issue: GitHub.