Skip to content

runai submit-mpi


Submit a Distributed Training (MPI) Run:AI Job for execution.


runai submit-mpi
    [--backoffLimit int] 
    [--cpu double] 
    [--cpu-limit double] 
    [--environment stringArray | -e stringArray] 
    [--gpu double | -g double] 
    [--image string | -i string] 
    [--job-name-prefix string]
    [--memory string] 
    [--memory-limit string] 
    [--name string]
    [--node-type string] 
    [--processes int] 
    [--pvc [StorageClassName]:Size:ContainerMountPath:[ro]]
    [--template string] 
    [--tty | -t]
    [--volume stringArray | -v stringArray] 

    [--loglevel string] 
    [--project string | -p string] 
    [--help | -h]

    -- [COMMAND] [ARGS...] [options]

Syntax notes:

  • Options with a value type of stringArray mean that you can add multiple values. You can either separate values with a comma or add the flag twice.


start an unattended mpi training Job of name dist1, based on Project team-a using a quickstart-distributed image:

runai submit-mpi --name dist1 --processes=2 -g 1 \

(see: distributed training Quickstart).


Aliases and Shortcuts


The name of the Job.


Mark this Job as Interactive. Interactive Jobs are not terminated automatically by the system.

--template string

Provide the name of a template. A template can provide default and mandatory values.

The prefix to use to automatically generate a Job name with an incremental index. When a Job name is omitted Run:AI will generate a Job name. The optional --job-name-prefix flag creates Job names with the provided prefix.

--always-pull-image stringArray (deprecated)

Deprecated. Please use image-pull-policy=always instead. When starting a container, always pull the image from the registry, even if the image is cached on the running node. This is useful when you are re-saving updates to the image using the same tag, but may incur a penalty of performance degradation on Job start.


Default is false. If set to true, wait for the Pod to start running. When the pod starts running, attach to the Pod. The flag is equivalent to the command runai attach.

The --attach flag also sets --tty and --stdin to true.


If set, overrides the image's entrypoint with the command supplied after '--'

Example: --command --args 10000

-e stringArray | --environment stringArray

Define environment variables to be set in the container. To set multiple values add the flag multiple times (-e BATCH_SIZE=50 -e LEARNING_RATE=0.2) or separate by a comma (-e BATCH_SIZE:50,LEARNING_RATE:0.2).

--image string | -i string

Image to use when creating the container for this Job

--image-pull-policy string

Pulling policy of the image When starting a container. Options are:

  • always (default): force image pulling to check whether local image already exists. If the image already exists locally and has the same digest, then the image will not be downloaded.

  • ifNotPresent: the image is pulled only if it is not already present locally.

  • never: the image is assumed to exist locally. No attempt is made to pull the image.

For more information see Kubernetes documentation.

--local-image (deprecated)

Deprecated. Please use image-pull-policy=never instead. Use a local image for this Job. A local image is an image which exists on all local servers of the Kubernetes Cluster.


Keep stdin open for the container(s) in the pod, even if nothing is attached.

-t, --tty

Allocate a pseudo-TTY

--working-dir string

Starts the container with the specified directory as the current directory.

Resource Allocation

--cpu double

CPU units to allocate for the Job (0.5, 1, .etc). The Job will receive at least this amount of CPU. Note that the Job will not be scheduled unless the system can guarantee this amount of CPUs to the Job.

--cpu-limit double

Limitations on the number of CPU consumed by the Job (0.5, 1, .etc). The system guarantees that this Job will not be able to consume more than this amount of GPUs.

--gpu double | -g double

Number of GPUs to allocate to the Job. The default is no allocated GPUs. the GPU value can be an integer or a fraction between 0 and 1.


Mount a large /dev/shm device. An shm is a shared file system mounted on RAM.

--memory string

CPU memory to allocate for this Job (1G, 20M, .etc). The Job will receive at least this amount of memory. Note that the Job will not be scheduled unless the system can guarantee this amount of memory to the Job.

--memory-limit string

CPU memory to allocate for this Job (1G, 20M, .etc). The system guarantees that this Job will not be able to consume more than this amount of memory. The Job will receive an error when trying to allocate more memory than this limit.


--pvc [Storage_Class_Name]:Size:Container_Mount_Path:[ro]

--pvc Pvc_Name:Container_Mount_Path:[ro]

Mount a persistent volume claim into a container.

The 2 syntax types of this command are mutually exclusive. You can either use the first or second form, but not a mixture of both.

Storage_Class_Name is a storage class name which can be obtained by running kubectl get This parameter may be omitted if there is a single storage class in the system, or you are using the default storage class.

Size is the volume size you want to allocate. See Kubernetes documentation for how to specify volume sizes

Container_Mount_Path. A path internal to the container where the storage will be mounted

Pvc_Name. The name of a pre-existing Persistent Volume Claim to mount into the container


--pvc :3Gi:/tmp/john:ro - Allocate 3GB from the default Storage class. Mount it to /tmp/john as read-only

--pvc my-storage:3Gi:/tmp/john:ro - Allocate 3GB from the my-storage storage class. Mount it to /tmp/john as read-only

--pvc :3Gi:/tmp/john - Allocate 3GB from the default storage class. Mount it to /tmp/john as read-write

--pvc my-pvc:/tmp/john - Use a Persistent Volume Claim named my-pvc. Mount it to /tmp/john as read-write

--pvc my-pvc-2:/tmp/john:ro - Use a Persistent Volume Claim named my-pvc-2. Mount it to /tmp/john as read-only

--volume stringArray | -v stringArray

Volume to mount into the container. Example -v /raid/public/john/data:/root/data:ro The flag may optionally be suffixed with :ro or :rw to mount the volumes in read-only or read-write mode, respectively.



Use the host's ipc namespace. Controls whether the pod containers can share the host IPC namespace. IPC (POSIX/SysV IPC) namespace provides separation of named shared memory segments, semaphores and message queues. Shared memory segments are used to accelerate inter-process communication at memory speed, rather than through pipes or through the network stack.

For further information see docker run reference documentation.


Use the host's networkstack inside the container. For further information see docker run referencedocumentation.

Job Lifecycle

--backoffLimit int

The number of times the Job will be retried before failing. The default is 6. This flag will only work with training workloads (when the --interactive flag is not specified).

--processes int

Number of distributed training processes. The default is 1.

Access Control


Create a temporary home directory for the user in the container. Data saved in this directory will not be saved when the container exits. The flag is set by default to true when the --run-as-user flag is used, and false if not.


Prevent the Job’s container and all launched processes from gaining additional privileges after the Job starts. Default is false. For more information see Privilege Escalation.


Run in the context of the current user running the Run:AI command rather than the root user. While the default container user is root (same as in Docker), this command allows you to submit a Job running under your Linux user. This would manifest itself in access to operating system resources, in the owner of new folders created under shared directories etc.


--node-type string

Allows defining specific Nodes (machines) or a group of Nodes on which the workload will run. To use this feature your administrator will need to label nodes as explained here: Limit a Workload to a Specific Node Group. This flag can be used in conjunction with Project-based affinity. In this case, the flag is used to refine the list of allowable node groups set in the Project. For more information see: Working with Projects.

Global Flags

--loglevel (string)

Set the logging level. One of: debug | info | warn | error (default "info")

--project | -p (string)

Specify the Project to which the command applies. Run:AI Projects are used by the scheduler to calculate resource eligibility. By default, commands apply to the default Project. To change the default Project use runai config project <project-name>.

--help | -h

Show help text.


The command will attempt to submit an mpi Job. You can follow up on the Job by running runai list jobs or runai describe job <job-name>.

See Also

Last update: December 1, 2020