abiosoft/colima · error

error setting up AI model runner: %w

Error message

error setting up AI model runner: %w

What it means

ProvisionRamalama streams a `set -e` install script into the VM (install ramalama, docker pull quay.io/ramalama/ramalama and ramalama-rag, chown /var/lib/ramalama, create the ~/.local/share/ramalama symlink) via guest.RunInteractive. Because of set -e, the first failing step aborts the script and surfaces here with the shell exit error wrapped in %w.

Source

Thrown at model/ramalama.go:177

  pipx upgrade ramalama
else
  pipx install ramalama
fi

# pull ramalama container images
docker pull quay.io/ramalama/ramalama
docker pull quay.io/ramalama/ramalama-rag

# fix ownership of persistent data dir and symlink to expected location
sudo chown -R $(id -u):$(id -g) /var/lib/ramalama
mkdir -p "$HOME/.local/share"
ln -sfn /var/lib/ramalama "$HOME/.local/share/ramalama"
`

	log.Println("installing AI model runner...")

	if err := guest.RunInteractive("sh", "-c", script); err != nil {
		return fmt.Errorf("error setting up AI model runner: %w", err)
	}

	log.Println("AI model runner installed")

	// mark as provisioned
	if err := store.Set(func(s *store.Store) {
		s.RamalamaProvisioned = true
	}); err != nil {
		return fmt.Errorf("error saving provisioning state: %w", err)
	}

	return nil
}

View on GitHub (pinned to c3a5f9184d)

Solutions

1. Identify the failing line by running the same script steps manually via colima ssh and watching where it stops
2. Verify VM connectivity to the script's endpoints: colima ssh -- curl -I https://quay.io
3. Check docker inside the VM: colima ssh -- docker info; restart colima if the daemon is wedged
4. Free or grow VM disk (colima stop && colima start --disk) — the two images need several GB

When it happens

Trigger: VM offline so the install or docker pull steps fail; docker daemon unresponsive inside the VM; quay.io unreachable or rate-limiting; /var/lib/ramalama absent so the chown fails; VM disk too small for the two images.

Common situations: First-ever colima model run behind a corporate VPN/proxy that blocks quay.io or the ramalama install host; default VM disk too small for the AI images.

Related errors

• failed to pull model %q: %w
• error creating systemd unit file: %w
• error reloading systemd daemon: %w
• error creating runtime disk: %w
• error listing models: %w

AI-assisted analysis of abiosoft/colima@c3a5f9184d (2026-08-15). Data as JSON: /api/errors/4370d4f87f891cbc. Report an issue: GitHub.