Velero has a plugin architecture that allows users to add their own custom functionality to Velero backups & restores without having to modify/recompile the core Velero binary. To add custom functionality, users simply create their own binary containing implementations of Velero’s plugin kinds (described below), plus a small amount of boilerplate code to expose the plugin implementations to Velero. This binary is added to a container image that serves as an init container for the Velero server pod and copies the binary into a shared emptyDir volume for the Velero server to access.
Multiple plugins, of any type, can be implemented in this binary.
A fully-functional sample plugin repository is provided to serve as a convenient starting point for plugin authors.
A plugin is identified by a prefix + name.
Note: Please don’t use
velero.io as the prefix for a plugin not supported by the Velero team. The prefix should help users identify the entity developing the plugin, so please use a prefix that identify yourself.
Whenever you define a Backup Storage Location or Volume Snapshot Location, this full name will be the value for the
apiVersion: velero.io/v1 kind: BackupStorageLocation spec: provider: oracle.io/oracle
apiVersion: velero.io/v1 kind: VolumeSnapshotLocation spec: provider: oracle.io/oracle
When naming your plugin, keep in mind that the full name needs to conform to these rules:
- example.io/azure - 184.108.40.206/5678 - example-with-dash.io/azure
You will need to give your plugin(s) the full name when registering them by calling the appropriate
RegisterX function: https://github.com/vmware-tanzu/velero/blob/0e0f357cef7cf15d4c1d291d3caafff2eeb69c1e/pkg/plugin/framework/server.go#L42-L60
Velero supports the following kinds of plugins:
Velero provides a
logger that can be used by plugins to log structured information to the main Velero server log or
per-backup/restore logs. It also passes a
--log-level flag to each plugin binary, whose value is the value of the same
flag from the main Velero process. This means that if you turn on debug logging for the Velero server via
plugins will also emit debug-level logs. See the
sample repository for an example of how to use the logger within your plugin.
Velero uses a ConfigMap-based convention for providing configuration to plugins. If your plugin needs to be configured at runtime, define a ConfigMap like the following:
apiVersion: v1 kind: ConfigMap metadata: # any name can be used; Velero uses the labels (below) # to identify it rather than the name name: my-plugin-config # must be in the namespace where the velero deployment # is running namespace: velero labels: # this value-less label identifies the ConfigMap as # config for a plugin (the built-in change storageclass # restore item action plugin) velero.io/plugin-config: "" # add a label whose key corresponds to the fully-qualified # plugin name (for example mydomain.io/my-plugin-name), and whose # value is the plugin type (BackupItemAction, RestoreItemAction, # ObjectStore, or VolumeSnapshotter) <fully-qualified-plugin-name>: <plugin-type> data: # add your configuration data here as key-value pairs
Then, in your plugin’s implementation, you can read this ConfigMap to fetch the necessary configuration. See the
restic restore action
for an example of this – in particular, the
Velero will pass any known features flags as a comma-separated list of strings to the
Once parsed into a
string, the features can then be registered using the
NewFeatureFlagSet function and queried with
Velero adds the
LD_LIBRARY_PATH into the list of environment variables to provide the convenience for plugins that requires C libraries/extensions in the runtime.