Skip to main content

Upgrade to Spaces

The Upbound migration tool is a CLI command that helps you update your existing Crossplane control plane to a managed Upbound Crossplane control plane in an Upbound Space.

To upgrade from Crossplane to Upbound, you must:

  1. Export your existing Crossplane control plane configuration/state into an archive file.
  2. Import the archive file into a control plane running in Upbound. The migration tool is available in the up CLI as up controlplane migration export and up controlplane migration import commands.

Prerequisites​

Before you begin, you must have the following:

  • The up CLI version 0.23.0 or later.

Migration process​

To upgrade an existing Crossplane control plane to a control plane in Upbound, do the following:

  1. Run the up controlplane migration export command to export your existing Crossplane control plane configuration/state into an archive file:

    up controlplane migration export --kubeconfig <path-to-source-kubeconfig> --output <path-to-archive-file>

    The command exports your existing Crossplane control plane configuration/state into an archive file.

note

By default, the export command doesn't make any changes to your existing Crossplane control plane state, leaving it intact. Use the --pause-before-export flag to pause the reconciliation on managed resources before exporting the archive file.

This is a safety mechanism to help ensure the control plane you upgrade state to doesn't assume ownership of resources before you're ready.

  1. Use the control plane create command to create a managed control plane in Upbound:

    up controlplane create my-controlplane

  2. Use up ctx to connect to the control plane created in the previous step:

    up ctx "<your-org>/<your-space>/<your-group>/my-controlplane"

    The command configures your local kubeconfig to connect to the control plane.

  3. Run the following command to import the archive file into the control plane:

    up controlplane migration import --input <path-to-archive-file>
note

By default, the import command leaves the control plane in an inactive state by pausing the reconciliation on managed resources which gives you an opportunity to review the imported configuration/state before activating the control plane. Use the --unpause-after-import flag to change the default behavior and activate the control plane immediately after importing the archive file.

  1. Review and validate the imported configuration/state. When you are ready, activate your managed control plane by running the following command:

    kubectl annotate managed --all crossplane.io/paused-

    At this point, you can delete the source Crossplane control plane.

CLI options​

Filtering​

The migration tool captures the state of a Control Plane. The only filtering supported is Kubernetes namespace and Kubernetes resource Type filtering.

You can exclude namespaces using the --exclude-namespaces CLI option. This can prevent the CLI from including unwanted resources in the export.

--exclude-namespaces=kube-system,kube-public,kube-node-lease,local-path-storage,...

# A list of specific namespaces to exclude from the export. Defaults to 'kube-system', 'kube-public','kube-node-lease', and 'local-path-storage'.

You can exclude Kubernetes Resource types by using the --exclude-resources CLI option:

--exclude-resources=EXCLUDE-RESOURCES,... 

# A list of resource types to exclude from the export in "resource.group" format. No resources are excluded by default.

For example, here's an example for excluding the CRDs installed by Crossplane functions (since they're not needed):

up controlplane migration export \
  --exclude-resources=gotemplates.gotemplating.fn.crossplane.io,kclinputs.template.fn.crossplane.io

After export, users can also change the archive file to only include necessary resources.

Export non-Crossplane resources​

Use the --include-extra-resources= CLI option to select other CRD types to include in the export.

Set the kubecontext​

Currently --context isn't supported in the migration CLI. You should be able to use the --kubeconfig CLI option to use a file that's set to the correct context. For example:

up controlplane migration export --kubeconfig

Use this in tandem with up ctx to export a control plane's kubeconfig:

up ctx --kubeconfig ~/.kube/config

# To list the current contet
up ctx . --kubeconfig ~/.kube/config

Export archive​

The migration CLI exports an archive upon successful completion. Below is an example export of a control plane that excludes several CRD types and skips the confirmation prompt. A file gets written to the working directory, unless you select another output file:

View the example export
$ up controlplane migration export
--exclude-resources=gotemplates.gotemplating.fn.crossplane.io,kclinputs.templat
e.fn.crossplane.io --yes
Exporting control plane state...
✓ Scanning control plane for types to export... 121 types found! 👀
✓ Exporting 121 Crossplane resources...60 resources exported! 📤
✓ Exporting 3 native resources...8 resources exported! 📤
✓ Archiving exported state... archived to "xp-state.tar.gz"! 📦

When an export occurs, a file named xp-state.tar.gz by default gets created in the working directory. You can unzip the file and all the contents of the export are all text YAML files.

  • Each CRD (for example vpcs.ec2.aws.upbound.io) gets its own directory which contains:
    • A metadata.yaml file that contains Kubernetes Object Metadata
    • A list of Kubernetes Categories the resource belongs to
  • A cluster directory that contains YAML manifests for all resources provisioned using the CRD.

Sample contents for a Cluster with a single XNetwork Composite from configuration-aws-network is show below:

View the example cluster content
├── compositionrevisions.apiextensions.crossplane.io
│ ├── cluster
│ │ ├── kcl.xnetworks.aws.platform.upbound.io-4ca6a8a.yaml
│ │ └── xnetworks.aws.platform.upbound.io-9859a34.yaml
│ └── metadata.yaml
├── configurations.pkg.crossplane.io
│ ├── cluster
│ │ └── configuration-aws-network.yaml
│ └── metadata.yaml
├── deploymentruntimeconfigs.pkg.crossplane.io
│ ├── cluster
│ │ └── default.yaml
│ └── metadata.yaml
├── export.yaml
├── functions.pkg.crossplane.io
│ ├── cluster
│ │ ├── crossplane-contrib-function-auto-ready.yaml
│ │ ├── crossplane-contrib-function-go-templating.yaml
│ │ └── crossplane-contrib-function-kcl.yaml
│ └── metadata.yaml
├── internetgateways.ec2.aws.upbound.io
│ ├── cluster
│ │ └── borrelli-backup-test-xgl4q.yaml
│ └── metadata.yaml
├── mainroutetableassociations.ec2.aws.upbound.io
│ ├── cluster
│ │ └── borrelli-backup-test-t2qh7.yaml
│ └── metadata.yaml
├── namespaces
│ └── cluster
│ ├── crossplane-system.yaml
│ ├── default.yaml
│ └── upbound-system.yaml
├── providerconfigs.aws.upbound.io
│ ├── cluster
│ │ └── default.yaml
│ └── metadata.yaml
├── providerconfigusages.aws.upbound.io
│ ├── cluster
│ │ ├── 0a2a3ec6-ef13-45f9-9cf0-63af7f4a6b6b.yaml
...redacted
│ │ └── f7092b0f-3a78-4bfe-82c8-57e5085a9b11.yaml
│ └── metadata.yaml
├── providers.pkg.crossplane.io
│ ├── cluster
│ │ ├── upbound-provider-aws-ec2.yaml
│ │ └── upbound-provider-family-aws.yaml
│ └── metadata.yaml
├── routes.ec2.aws.upbound.io
│ ├── cluster
│ │ └── borrelli-backup-test-dt9cj.yaml
│ └── metadata.yaml
├── routetableassociations.ec2.aws.upbound.io
│ ├── cluster
│ │ ├── borrelli-backup-test-mr2sd.yaml
│ │ ├── borrelli-backup-test-ngq5h.yaml
│ │ ├── borrelli-backup-test-nrkgg.yaml
│ │ └── borrelli-backup-test-wq752.yaml
│ └── metadata.yaml
├── routetables.ec2.aws.upbound.io
│ ├── cluster
│ │ └── borrelli-backup-test-dv4mb.yaml
│ └── metadata.yaml
├── secrets
│ └── namespaces
│ ├── crossplane-system
│ │ ├── cert-token-signing-gateway-pub.yaml
│ │ ├── mxp-hostcluster-certs.yaml
│ │ ├── package-pull-secret.yaml
│ │ └── xgql-tls.yaml
│ └── upbound-system
│ └── aws-creds.yaml
├── securitygrouprules.ec2.aws.upbound.io
│ ├── cluster
│ │ ├── borrelli-backup-test-472f4.yaml
│ │ └── borrelli-backup-test-qftmw.yaml
│ └── metadata.yaml
├── securitygroups.ec2.aws.upbound.io
│ ├── cluster
│ │ └── borrelli-backup-test-w5jch.yaml
│ └── metadata.yaml
├── storeconfigs.secrets.crossplane.io
│ ├── cluster
│ │ └── default.yaml
│ └── metadata.yaml
├── subnets.ec2.aws.upbound.io
│ ├── cluster
│ │ ├── borrelli-backup-test-8btj6.yaml
│ │ ├── borrelli-backup-test-gbmrm.yaml
│ │ ├── borrelli-backup-test-m7kh7.yaml
│ │ └── borrelli-backup-test-nttt5.yaml
│ └── metadata.yaml
├── vpcs.ec2.aws.upbound.io
│ ├── cluster
│ │ └── borrelli-backup-test-7hwgh.yaml
│ └── metadata.yaml
└── xnetworks.aws.platform.upbound.io
├── cluster
│ └── borrelli-backup-test.yaml
└── metadata.yaml
43 directories, 87 files

The export.yaml file contains metadata about the export, including the configuration of the export, Crossplane information, and what's included in the export bundle.

View the export
version: v1alpha1
exportedAt: 2025-01-06T17:39:53.173222Z
options:
  excludedNamespaces:
    - kube-system
    - kube-public
    - kube-node-lease
    - local-path-storage
  includedResources:
    - namespaces
    - configmaps
    - secrets
  excludedResources:
    - gotemplates.gotemplating.fn.crossplane.io
    - kclinputs.template.fn.crossplane.io
crossplane:
  distribution: universal-crossplane
  namespace: crossplane-system
  version: 1.17.3-up.1
  featureFlags:
    - --enable-provider-identity
    - --enable-environment-configs
    - --enable-composition-functions
    - --enable-usages
stats:
  total: 68
  nativeResources:
    configmaps: 0
    namespaces: 3
    secrets: 5
  customResources:
    amicopies.ec2.aws.upbound.io: 0
    amilaunchpermissions.ec2.aws.upbound.io: 0
    amis.ec2.aws.upbound.io: 0
    availabilityzonegroups.ec2.aws.upbound.io: 0
    capacityreservations.ec2.aws.upbound.io: 0
    carriergateways.ec2.aws.upbound.io: 0
    compositeresourcedefinitions.apiextensions.crossplane.io: 0
    compositionrevisions.apiextensions.crossplane.io: 2
    compositions.apiextensions.crossplane.io: 0
    configurationrevisions.pkg.crossplane.io: 0
    configurations.pkg.crossplane.io: 1
...redacted

Skipped resources​

Along with to the resources excluded via CLI options, the following resources aren't included in the backup:

  • The kube-root-ca.crt ConfigMap, since this is cluster-specific
  • Resources directly managed via Helm (ArgoCD's helm implementation, which templates Helm resources and then applies them, get included in the backup). The migration creates the exclusion list by looking for:
    • Any Resource with the label "app.kubernetes.io/managed-by" == "Helm"
    • Kubernetes Secrets with the label prefix helm.sh/release. For example, helm.sh/release.v1
  • Resources installed via a Crossplane package. These have an ownerReference with a prefix pkg.crossplane.io. The expectation is that during import, the Crossplane Package Manager bears responsibility for installing the resources.
  • Crossplane Locks: Any Lock.pkg.crossplane.io resource isn't included in the export.

Restore​

The following is an example of a successful import run. At the end of the import, all Managed Resources are in a paused state.

View the migration import
$ up controlplane migration import
Importing control plane state...
✓ Reading state from the archive... Done! 👀
✓ Importing base resources... 18 resources imported! 📥
✓ Waiting for XRDs... Established! ⏳
✓ Waiting for Packages... Installed and Healthy! ⏳
✓ Importing remaining resources... 50 resources imported! 📥
✓ Finalizing import... Done! 🎉

Your scenario may involve migrating resources which already exist through other automation on the platform. When executing an import in these circumstances, the importer applies the new manifests to the cluster. If the resource already exists, the restore sets fields to what's in the backup.

The importer restores all resources in the export archive. Managed Resources get imported with the crossplane.io/paused: "true" annotation set. Use the --unpause-after-import CLI argument to automatically un-pause resources that got paused during backup, or remove the annotation manually.

Restore order​

The importer restores based on Kubernetes types. The restore order doesn't include parent/child relationships.

Because Crossplane Composites create new Managed Resources if not present on the cluster, all Claims, Composites and Managed Resources get imported in a paused state. You can un-pause them after the restore completes.

The first step of import is installing Base Resources into the cluster. These resources (such has packages and XRDs) must be ready before proceeding with the import. Base Resources are:

  • Kubernetes Resources
    • ConfigMaps
    • Namespaces
    • Secrets
  • Crossplane Resources
    • ControllerConfigs: controllerconfigs.pkg.crossplane.io
    • DeploymentRuntimeConfigs: deploymentruntimeconfigs.pkg.crossplane.io
    • StoreConfigs: storeconfigs.secrets.crossplane.io
  • Crossplane Packages
    • Providers: providers.pkg.crossplane.io
    • Functions: functions.pkg.crossplane.io
    • Configurations: configurations.pkg.crossplane.io

Restore waits for the base resources to be Ready before moving on to the next step. Next, restore walks through the archive and restores all the manifests present.

During import, the crossplane.io/paused annotation gets added to Managed Resources, Claims and Composites.

To manually un-pause managed resources after an import, remove the annotation by running:

kubectl annotate managed --all crossplane.io/paused-

You can also run import again with the --unpause-after-import flag to remove the annotations.

up controlplane migration import --unpause-after-import

Restoring resource status​

The importer applies the status of all resources during import. The importer determines if the CRD version has a status field defined based on the stored CRD version.