github.com/anth0d/nomad@v0.0.0-20221214183521-ae3a0a2cad06/website/content/docs/concepts/plugins/devices.mdx (about)

     1  ---
     2  layout: docs
     3  page_title: Device Plugins
     4  description: Learn how to author a Nomad device plugin.
     5  ---
     6  
     7  # Devices
     8  
     9  Nomad has built-in support for scheduling compute resources such as CPU, memory,
    10  and networking. Nomad device plugins are used to support scheduling tasks with
    11  other devices, such as GPUs. They are responsible for fingerprinting these
    12  devices and working with the Nomad client to make them available to assigned
    13  tasks.
    14  
    15  For a real world example of a Nomad device plugin implementation, see the [Nvidia
    16  GPU plugin](https://github.com/hashicorp/nomad-device-nvidia).
    17  
    18  ## Authoring Device Plugins
    19  
    20  Authoring a device plugin in Nomad consists of implementing the
    21  [DevicePlugin][deviceplugin] interface alongside
    22  a main package to launch the plugin.
    23  
    24  The [device plugin skeleton project][skeletonproject] exists to help bootstrap
    25  the development of new device plugins. It provides most of the boilerplate
    26  necessary for a device plugin, along with detailed comments.
    27  
    28  ### Lifecycle and State
    29  
    30  A device plugin is long-lived. Nomad will ensure that one instance of the plugin is
    31  running. If the plugin crashes or otherwise terminates, Nomad will launch another
    32  instance of it.
    33  
    34  However, unlike [task drivers](/docs/concepts/plugins/task-drivers), device plugins do not currently
    35  have an interface for persisting state to the Nomad client. Instead, the device
    36  plugin API emphasizes fingerprinting devices and reporting their status. After
    37  helping to provision a task with a scheduled device, a device plugin does not
    38  have any responsibility (or ability) to monitor the task.
    39  
    40  ## Device Plugin API
    41  
    42  The [base plugin][baseplugin] must be implemented in addition to the following
    43  functions.
    44  
    45  ### `Fingerprint(context.Context) (<-chan *FingerprintResponse, error)`
    46  
    47  The `Fingerprint` [function][fingerprintfn] is called by the client when the plugin is started.
    48  It allows the plugin to provide Nomad with a list of discovered devices, along with their
    49  attributes, for the purpose of scheduling workloads using devices.
    50  The channel returned should immediately send an initial
    51  [`FingerprintResponse`][fingerprintresponse], then send periodic updates at
    52  an appropriate interval until the context is canceled.
    53  
    54  Each fingerprint response consists of either an error or a list of device groups.
    55  A device group is a list of detected devices that are identical for the purpose of
    56  scheduling; that is, they will have identical attributes.
    57  
    58  ### `Stats(context.Context, time.Duration) (<-chan *StatsResponse, error)`
    59  
    60  The `Stats` [function][statsfn] returns a channel on which the plugin should
    61  emit device statistics, at the specified interval, until either an error is
    62  encountered or the specified context is cancelled. The `StatsReponse` object
    63  allows [dimensioned][dimensioned] statistics to be returned for each device in a device group.
    64  
    65  ### `Reserve(deviceIDs []string) (*ContainerReservation, error)`
    66  
    67  The `Reserve` [function][reservefn] accepts a list of device IDs and returns the information
    68  necessary for the client to make those devices available to a task. Currently,
    69  the `ContainerReservation` object allows the plugin to specify environment
    70  variables for the task, as well as a list of host devices and files to be mounted
    71  into the task's filesystem. Any orchestration required to prepare the device for
    72  use should also be performed in this function.
    73  
    74  [deviceplugin]: https://github.com/hashicorp/nomad/blob/v0.9.0/plugins/device/device.go#L20-L33
    75  [baseplugin]: /docs/concepts/plugins/base
    76  [skeletonproject]: https://github.com/hashicorp/nomad-skeleton-device-plugin
    77  [fingerprintresponse]: https://github.com/hashicorp/nomad/blob/v0.9.0/plugins/device/device.go#L37-L43
    78  [fingerprintfn]: https://github.com/hashicorp/nomad-skeleton-device-plugin/blob/v0.1.0/device/device.go#L159-L165
    79  [statsfn]: https://github.com/hashicorp/nomad-skeleton-device-plugin/blob/v0.1.0/device/device.go#L169-L176
    80  [reservefn]: https://github.com/hashicorp/nomad-skeleton-device-plugin/blob/v0.1.0/device/device.go#L189-L245
    81  [dimensioned]: https://github.com/hashicorp/nomad/blob/v0.9.0/plugins/shared/structs/stats.go#L33-L34