kubernetes/kops · error

Error reading systemd file %q: %v

Error message

Error reading systemd file %q: %v

What it means

Service.Find reads the unit file at <systemdSystemPath>/<name>. If the read fails for any reason other than the file not existing (permissions, I/O error, path is a directory), the task cannot compare actual vs desired state and returns this error instead of treating the service as absent.

Source

Thrown at upup/pkg/fi/nodeup/nodetasks/service.go:203

func (e *InstallService) Find(_ *fi.InstallContext) (*InstallService, error) {
	actual, err := e.Service.Find(nil)
	if actual == nil || err != nil {
		return nil, err
	}
	return &InstallService{*actual}, nil
}
func (e *Service) Find(_ *fi.NodeupContext) (*Service, error) {
	systemdSystemPath, err := e.systemdSystemPath()
	if err != nil {
		return nil, err
	}

	servicePath := path.Join(systemdSystemPath, e.Name)

	d, err := os.ReadFile(servicePath)
	if err != nil {
		if !os.IsNotExist(err) {
			return nil, fmt.Errorf("Error reading systemd file %q: %v", servicePath, err)
		}

		// Not found
		return &Service{
			Name:       e.Name,
			Definition: nil,
			Running:    new(false),
		}, nil
	}

	actual := &Service{
		Name:       e.Name,
		Definition: new(string(d)),

		// Avoid spurious changes
		ManageState:  e.ManageState,
		SmartRestart: e.SmartRestart,
	}

View on GitHub (pinned to 4c8573c808)

Solutions

1. Check permissions/ownership of the unit file: ls -l /etc/systemd/system/<name>; make it readable (chmod 0644) by the nodeup user
2. If the path is a directory or stale file, remove it and re-run nodeup
3. Check dmesg / filesystem health if the error is I/O related
4. Re-run nodeup as root, which is the normal invocation for nodeup

Example fix

# before
$ ls -l /etc/systemd/system/kops.service
-rw------- 1 root root ...  # unreadable by nodeup user
# after
$ chmod 0644 /etc/systemd/system/kops.service

// Pre-check on the node:
ls -l /etc/systemd/system/<unit> 2>/dev/null && sudo cat /etc/systemd/system/<unit> >/dev/null && echo readable || echo 'unreadable or missing'

if _, err := svc.Find(ctx); err != nil {
	if strings.Contains(err.Error(), "Error reading systemd file") {
		// check perms / fs health before retrying
		return fmt.Errorf("unit file unreadable: %w", err)
	}
	return err
}

When it happens

Trigger: os.ReadFile on path.Join(systemdSystemPath, e.Name) fails with a non-ENOENT error: unit file exists but is unreadable (wrong permissions), the path is a directory, or the filesystem returns EIO/EBUSY.

Common situations: Unit files created by another tool with restrictive modes (e.g. 0600 root-only while nodeup runs as non-root), leftover directories where a unit file is expected, or corrupted node disks producing I/O errors on /etc/systemd/system.

Related errors

• error writing systemd service file: %v
• failed to update /etc/hosts: %w
• creating directories %q: %w
• unable to read snippet: %s, error: %s
• unable to read template: %s, error: %s

AI-assisted analysis of kubernetes/kops@4c8573c808 (2026-09-05). Data as JSON: /api/errors/3f8338160ce2c35a. Report an issue: GitHub.