Device Plugins
Starting in version 1.8, Kubernetes provides a device plugin framework for vendors to advertise their resources to the kubelet without changing Kubernetes core code. Instead of writing custom Kubernetes code, vendors can implement a device plugin that can be deployed manually or as a DaemonSet. The targeted devices include GPUs, High-performance NICs, FPGAs, InfiniBand, and other similar computing resources that may require vendor specific initialization and setup.
Device plugin registration
The device plugins feature is gated by the DevicePlugins feature gate which
is disabled by default before 1.10. When the device plugins feature is enabled,
the kubelet exports a Registration gRPC service:
service Registration {
rpc Register(RegisterRequest) returns (Empty) {}
}A device plugin can register itself with the kubelet through this gRPC service. During the registration, the device plugin needs to send:
- The name of its Unix socket.
- The Device Plugin API version against which it was built.
- The
ResourceNameit wants to advertise. HereResourceNameneeds to follow the extended resource naming scheme asvendor-domain/resource. For example, an Nvidia GPU is advertised asnvidia.com/gpu.
Following a successful registration, the device plugin sends the kubelet the
list of devices it manages, and the kubelet is then in charge of advertising those
resources to the API server as part of the kubelet node status update.
For example, after a device plugin registers vendor-domain/foo with the kubelet
and reports two healthy devices on a node, the node status is updated
to advertise 2 vendor-domain/foo.
Then, users can request devices in a Container specification as they request other types of resources, with the following limitations:
- Extended resources are only supported as integer resources and cannot be overcommitted.
- Devices cannot be shared among Containers.
Suppose a Kubernetes cluster is running a device plugin that advertises resource vendor-domain/resource
on certain nodes, here is an example user pod requesting this resource:
apiVersion: v1
kind: Pod
metadata:
name: demo-pod
spec:
containers:
- name: demo-container-1
image: k8s.gcr.io/pause:2.0
resources:
limits:
vendor-domain/resource: 2 # requesting 2 vendor-domain/resourceDevice plugin implementation
The general workflow of a device plugin includes the following steps:
Initialization. During this phase, the device plugin performs vendor specific initialization and setup to make sure the devices are in a ready state.
The plugin starts a gRPC service, with a Unix socket under host path
/var/lib/kubelet/device-plugins/, that implements the following interfaces:
service DevicePlugin {
// ListAndWatch returns a stream of List of Devices
// Whenever a Device state change or a Device disappears, ListAndWatch
// returns the new list
rpc ListAndWatch(Empty) returns (stream ListAndWatchResponse) {}
// Allocate is called during container creation so that the Device
// Plugin can run device specific operations and instruct Kubelet
// of the steps to make the Device available in the container
rpc Allocate(AllocateRequest) returns (AllocateResponse) {}
}The plugin registers itself with the kubelet through the Unix socket at host path
/var/lib/kubelet/device-plugins/kubelet.sock.After successfully registering itself, the device plugin runs in serving mode, during which it keeps monitoring device health and reports back to the kubelet upon any device state changes. It is also responsible for serving
AllocategRPC requests. DuringAllocate, the device plugin may do device-specific preparation; for example, GPU cleanup or QRNG initialization. If the operations succeed, the device plugin returns anAllocateResponsethat contains container runtime configurations for accessing the allocated devices. The kubelet passes this information to the container runtime.
A device plugin is expected to detect kubelet restarts and re-register itself with the new
kubelet instance. In the current implementation, a new kubelet instance deletes all the existing Unix sockets
under /var/lib/kubelet/device-plugins when it starts. A device plugin can monitor the deletion
of its Unix socket and re-register itself upon such an event.
Device plugin deployment
A device plugin can be deployed manually or as a DaemonSet. Being deployed as a DaemonSet has
the benefit that Kubernetes can restart the device plugin if it fails.
Otherwise, an extra mechanism is needed to recover from device plugin failures.
The canonical directory /var/lib/kubelet/device-plugins requires privileged access,
so a device plugin must run in a privileged security context.
If a device plugin is running as a DaemonSet, /var/lib/kubelet/device-plugins
must be mounted as a
Volume
in the plugin’s
PodSpec.
Kubernetes device plugin support is still in alpha. As development continues, its API version can change in incompatible ways. We recommend that device plugin developers do the following:
- Watch for changes in future releases.
- Support multiple versions of the device plugin API for backward/forward compatibility.
If you enable the DevicePlugins feature and run device plugins on nodes that need to be upgraded to a Kubernetes release with a newer device plugin API version, upgrade your device plugins to support both versions before upgrading these nodes to ensure the continuous functioning of the device allocations during the upgrade.
Examples
For examples of device plugin implementations, see:
- The official NVIDIA GPU device plugin
- it requires using nvidia-docker 2.0 which allows you to run GPU enabled docker containers
- The NVIDIA GPU device plugin for COS base OS.
- The RDMA device plugin
- The Solarflare device plugin
- The AMD GPU device plugin
- The SRIOV Network device plugin
FEATURE STATE:
Kubernetes v1.10
beta
This feature is currently in a beta state, meaning:
- The version names contain beta (e.g. v2beta3).
- Code is well tested. Enabling the feature is considered safe. Enabled by default.
- Support for the overall feature will not be dropped, though details may change.
- The schema and/or semantics of objects may change in incompatible ways in a subsequent beta or stable release. When this happens, we will provide instructions for migrating to the next version. This may require deleting, editing, and re-creating API objects. The editing process may require some thought. This may require downtime for applications that rely on the feature.
- Recommended for only non-business-critical uses because of potential for incompatible changes in subsequent releases. If you have multiple clusters that can be upgraded independently, you may be able to relax this restriction.
- Please do try our beta features and give feedback on them! After they exit beta, it may not be practical for us to make more changes.