kubernetes/kops · error

cannot update configuration file %s: does not exist

Error message

cannot update configuration file %s: does not exist

What it means

writeConfig() was invoked with vfs.WriteOptionOnlyIfExists (an update path) but a pre-flight ReadFile confirmed the configuration file does not exist in the state store. Updates are only allowed on existing objects, so this guards against silently creating a file via update.

Source

Thrown at pkg/client/simple/vfsclientset/commonvfs.go:149

	return object, nil
}

func (c *VFSClientBase) writeConfig(ctx context.Context, cluster *kops.Cluster, configPath vfs.Path, o runtime.Object, writeOptions ...vfs.WriteOption) error {
	data, err := c.serialize(o)
	if err != nil {
		return fmt.Errorf("error marshaling object: %v", err)
	}

	create := false
	for _, writeOption := range writeOptions {
		switch writeOption {
		case vfs.WriteOptionCreate:
			create = true
		case vfs.WriteOptionOnlyIfExists:
			_, err = configPath.ReadFile(ctx)
			if err != nil {
				if os.IsNotExist(err) {
					return fmt.Errorf("cannot update configuration file %s: does not exist", configPath)
				}
				return fmt.Errorf("error checking if configuration file %s exists already: %v", configPath, err)
			}
		default:
			return fmt.Errorf("unknown write option: %q", writeOption)
		}
	}

	acl, err := acls.GetACL(ctx, configPath, cluster)
	if err != nil {
		return err
	}

	rs := bytes.NewReader(data)
	if create {
		err = configPath.CreateFile(ctx, rs, acl)
	} else {
		err = configPath.WriteFile(ctx, rs, acl)

View on GitHub (pinned to 4c8573c808)

Solutions

1. Verify the cluster/object name spelling and the KOPS_STATE_STORE bucket are correct.
2. Check the file exists: `aws s3 ls $KOPS_STATE_STORE/<name>/cluster/spec` (or equivalent path).
3. Create the object first with Create instead of Update if it genuinely doesn't exist.
4. Restore from backup if the state store object was deleted unintentionally.

Example fix

// before
kops update cluster --name typo.example.com
// after
kops get clusters  # list actual cluster names
kops update cluster --name prod.example.com

existing, err := client.Find(ctx, name)
if err != nil { return err }
if existing == nil {
    return fmt.Errorf("object %q does not exist in state store; use Create instead of Update", name)
}

err := client.Update(ctx, obj)
if err != nil && strings.Contains(err.Error(), "does not exist") {
    // create it first or correct the name/state store
}

When it happens

Trigger: Update()/update() on a vfs clientset where the object file at basePath.Join(name) is missing — updating a cluster that was never created or was deleted.

Common situations: Running `kops update cluster` after `kops delete cluster` or with a mistyped cluster name; wrong KOPS_STATE_STORE so the object appears missing; racing deletion and update operations.

Related errors

• error reading full cluster spec for %q: %v
• error replacing cluster: %v
• reading kops-channels manifest %s: %w
• error parsing ConfigStore.Base %q: %v
• error reading state store: %v

AI-assisted analysis of kubernetes/kops@4c8573c808 (2026-09-05). Data as JSON: /api/errors/1f7d4c6cbfcc77c1. Report an issue: GitHub.