It was discovered during Velero 1.6.3 upgrade testing that Velero was
restoring `APIService` objects for APIs that are no longer being served
by Kubernetes 1.22. If these items were restored, it would break the
behaviour of discovery within the cluster.
This change introduces a new RestoreItemAction plugin that skips the
restore of any `APIService` object which is managed by Kubernetes such
as those for built-in APIs or CRDs. The `APIService`s for these will be
created when the Kubernetes API server starts or when new CRDs are
registered. These objects are identified by looking for the
`kube-aggregator.kubernetes.io/automanaged` label.
Signed-off-by: Bridget McErlean <bmcerlean@vmware.com>
* Switch to backing up v1beta1 CRDs from API server
Instead of simply switching out the APIVersion string on a v1
CustomResourceDefinition object, re-download the object from the API
server entirely to get the correct fields.
This should fix validation errors upon restore.
Signed-off-by: Nolan Brubaker <brubakern@vmware.com>
* Fix existing tests
Signed-off-by: Nolan Brubaker <brubakern@vmware.com>
* Add full example CRDs to automated tests
Signed-off-by: Nolan Brubaker <brubakern@vmware.com>
* Move beta CRD lookup into helper function
Signed-off-by: Nolan Brubaker <brubakern@vmware.com>
* Add case for preserveUnknownFields CRDs
Signed-off-by: Nolan Brubaker <brubakern@vmware.com>
* Add PreserveUnknownFields case and refactor execute
Signed-off-by: Nolan Brubaker <brubakern@vmware.com>
* Add older prometheus CRD test cases
Signed-off-by: Nolan Brubaker <brubakern@vmware.com>
* Add changelog
Signed-off-by: Nolan Brubaker <brubakern@vmware.com>
* Account for possible missing schemas on v1 CRDs
If a v1beta1 CRD without a Schema was submitted to a Kubernets v1.16
cluster, then Kubernetes will server it back as a v1 CRD without a
schema.
However, when Velero tries to restore this document, the request will be
rejected as a v1 CRD must have a schema.
This commit has some defensive coding on the restore side, as well as
potential fixes on the backup side for getting around this.
Signed-off-by: Nolan Brubaker <brubakern@vmware.com>
* Back up nonstructural CRDs as v1beta1
Signed-off-by: Nolan Brubaker <brubakern@vmware.com>
* Add tests for remapping plugin
Signed-off-by: Nolan Brubaker <brubakern@vmware.com>
* Add builders for v1 CRDs
Signed-off-by: Nolan Brubaker <brubakern@vmware.com>
* Address review feedback
Signed-off-by: Nolan Brubaker <brubakern@vmware.com>
* Remove extraneous log message
Signed-off-by: Nolan Brubaker <brubakern@vmware.com>
* Add changelog
Signed-off-by: Nolan Brubaker <brubakern@vmware.com>
* update import paths to github.com/vmware-tanzu/...
Signed-off-by: Steve Kriss <krisss@vmware.com>
* update other GH org refs to vmware-tanzu
Signed-off-by: Steve Kriss <krisss@vmware.com>
* site and docs: update GH org to vmware-tanzu
Signed-off-by: Steve Kriss <krisss@vmware.com>
* update travis badge links on docs readmes
Signed-off-by: Steve Kriss <krisss@vmware.com>
* add restore item action to change PVC/PV storage class name
Signed-off-by: Steve Kriss <krisss@vmware.com>
* code review
Signed-off-by: Steve Kriss <krisss@vmware.com>
* change existing plugin names to lowercase/hyphenated
Signed-off-by: Steve Kriss <krisss@vmware.com>
* add validation for new storage class name
Signed-off-by: Steve Kriss <krisss@vmware.com>
* add test cases
Signed-off-by: Steve Kriss <krisss@vmware.com>
* changelog
Signed-off-by: Steve Kriss <krisss@vmware.com>
* fix imports
Signed-off-by: Steve Kriss <krisss@vmware.com>
* update plugin names to be more consistent
Signed-off-by: Steve Kriss <krisss@vmware.com>
* update unit tests to use pkg/test object constructors
Signed-off-by: Steve Kriss <krisss@vmware.com>
Instead of only removing the default token from a service account when
it already exists in the cluster, always remove it. If the service
account already exists, continue to do the merging logic.
Signed-off-by: Andy Goldstein <andy.goldstein@gmail.com>
Refactor plugin management:
- support multiple plugins per executable
- support restarting a plugin process in the event it terminates
- simplify plugin lifecycle management by using separate managers for
each scope (server vs backup vs restore)
Signed-off-by: Andy Goldstein <andy.goldstein@gmail.com>
