kubernetes/kops · error

Error creating instance: %v

Error message

Error creating instance: %v

What it means

This error wraps any failure returned by the Nova compute API when kOps' RenderOpenstack tries to boot a new server (Cloud.CreateInstance) with a keypair extension, scheduler hint (server group), and pre-created port. It is a pass-through wrapper: the underlying gophercloud/OpenStack error (auth, quota, flavor/image, network port, server-group policy, etc.) is appended verbatim. Because the create call is the core of instance provisioning, any non-2xx from POST /servers surfaces here.

Source

Thrown at upup/pkg/fi/cloudup/openstacktasks/instance.go:381

			opt.UserData = bytes
		}
		if e.AvailabilityZone != nil {
			opt.AvailabilityZone = fi.ValueOf(e.AvailabilityZone)
		}
		if opt, err = includeBootVolumeOptions(t, e, opt); err != nil {
			return err
		}

		keyext := keypairs.CreateOptsExt{
			CreateOptsBuilder: opt,
			KeyName:           openstackKeyPairName(fi.ValueOf(e.SSHKey)),
		}

		schedulerHints := servers.SchedulerHintOpts{Group: *e.ServerGroup.ID}

		v, err := t.Cloud.CreateInstance(keyext, schedulerHints, fi.ValueOf(e.Port.ID))
		if err != nil {
			return fmt.Errorf("Error creating instance: %v", err)
		}
		e.ID = new(v.ID)

		if e.FloatingIP != nil {
			err = associateFloatingIP(t, e)
			if err != nil {
				return err
			}
		}

		klog.V(2).Infof("Creating a new Openstack instance, id=%s", v.ID)

		return nil
	}
	if changes.Port != nil {
		_, err := attachinterfaces.Create(context.TODO(), cloud.ComputeClient(), fi.ValueOf(e.ID), attachinterfaces.CreateOpts{
			PortID: fi.ValueOf(changes.Port.ID),
		}).Extract()

View on GitHub (pinned to 4c8573c808)

Solutions

1. Read the wrapped inner error to identify the Nova failure (quota vs not-found vs auth) and fix that specific cause first
2. Verify the image and flavor names in the cluster spec still exist: `openstack image list`, `openstack flavor list`
3. Check project quotas: `openstack quota show` and raise instance/cores/RAM limits if exceeded
4. Validate the port and network exist: `openstack port show <port-id>`; rerun `kops update cluster` to recreate stale ports
5. Confirm OS_* environment variables / clouds.yaml credentials and the correct region with `openstack server list`

Example fix

// before (config referencing missing image, error surfaces at create)
image: ubuntu-18.04-x86_64
// after (use an image that exists in the target region)
image: ubuntu-22.04-x86_64  # confirmed via `openstack image list`

// pre-flight: validate image, flavor, port, and quota before apply
func validateInstanceInputs(imageName, flavorName, portID string) error {
	if _, err := images.Get(context.TODO(), imageClient, imageName).Extract(); err != nil {
		return fmt.Errorf("image %q not found: %w", imageName, err)
	}
	if _, err := flavors.Get(context.TODO(), computeClient, flavorName).Extract(); err != nil {
		return fmt.Errorf("flavor %q not found: %w", flavorName, err)
	}
	if _, err := ports.Get(context.TODO(), networkClient, portID).Extract(); err != nil {
		return fmt.Errorf("port %q not found: %w", portID, err)
	}
	return nil
}

func isNotFoundErr(err error) bool {
	var _, nf gophercloud.ErrDefault404
	return errors.As(err, &nf) || strings.Contains(err.Error(), "could not be found")
}

_, err := t.Cloud.CreateInstance(keyext, schedulerHints, fi.ValueOf(e.Port.ID))
if err != nil {
	switch {
	case strings.Contains(err.Error(), "Quota"):
		return fmt.Errorf("quota exceeded, request limit raise: %w", err)
	case isNotFoundErr(err):
		return fmt.Errorf("referenced image/flavor/port missing: %w", err)
	default:
		return fmt.Errorf("Error creating instance: %w", err)
	}
}

When it happens

Trigger: POST /v2.1/{project}/servers fails during `kops update cluster` on OpenStack: invalid or missing imageRef/flavorRef, quota exceeded (instances, cores, RAM), port ID not found or in wrong network, server group ID invalid, keypair name not found, AvailabilityZone invalid, boot-from-volume block-device mapping rejected, or Nova API outage/401.

Common situations: Wrong cloud/OS_* credentials or expired token; cluster config referencing an image or flavor name that no longer exists in the region; tenant quota exhausted after scaling; port created by a previous failed apply is stale/deleted; Octavia/Nova version mismatch rejecting the keypairs CreateOptsExt; region endpoint misconfigured.

Related errors

• error building nova client: %v
• no decernable storage availability zone could be mapped to c
• error building nova client: %w
• could not delete instance %q: %v
• could not list server groups %v

AI-assisted analysis of kubernetes/kops@4c8573c808 (2026-09-05). Data as JSON: /api/errors/a7c05fe81117b49d. Report an issue: GitHub.