Skip to content

Submit a Run:ai Job via YAML


Researcher Kubernetes API is deprecated. See Cluster API for its replacement.

You can use YAML files to submit jobs directly to Kubernetes. A frequent scenario for using the Kubernetes YAML syntax to submit Jobs is integrations. Researchers may already be working with an existing system that submits Jobs, and want to continue working with the same system.


We differentiate between three types of Workloads:

  • Train workloads. Train workloads are characterized by a deep learning session that has a start and an end. A Training session can take anywhere from a few minutes to a couple of weeks. It can be interrupted in the middle and later restored. Training workloads typically utilize large percentages of GPU computing power and memory.
  • Build workloads. Build workloads are interactive. They are used by data scientists to write machine learning code and test it against subsets of the data. Build workloads typically do not maximize usage of the GPU.
  • Inference workloads. Inference workloads are used for serving models in production. For details on how to submit Inference workloads via YAML see here.

The internal Kubernetes implementation of a Run:ai Job is a CRD (Customer Resource) named RunaiJob which is similar to a Kubernetes Job.

Run:ai extends the Kubernetes Scheduler. A Kubernetes Scheduler is the software that determines which workload to start on which node. Run:ai provides a custom scheduler named runai-scheduler.

The Run:ai scheduler schedules computing resources by associating Workloads with Run:ai Projects:

  • A Project is assigned with a GPU quota through the Run:ai Run:ai User Interface.
  • A workload must be associated with a Project name and will receive resources according to the defined quota for the Project and the currently running Workloads

Internally, Run:ai Projects are implemented as Kubernetes namespaces. The scripts below assume that the code is being run after the relevant namespace has been set.

Submit Workloads

  • <JOB-NAME>. The name of the Job.
  • <IMAGE-NAME>. The name of the docker image to use. Example:
  • <USER-NAME>. The name of the user submitting the Job. The name is used for display purposes only when Run:ai is installed in an unauthenticated mode.
  • <REQUESTED-GPUs>. An integer number of GPUs you request to be allocated for the Job. Examples: 1, 2.
  • <NAMESAPCE>. The name of the Project's namespace. This is usually runai-<PROJECT-NAME>.

Regular Jobs

Copy the following into a file and change the parameters:

kind: RunaiJob (* see note below)
  name: <JOB-NAME>
  namespace: <NAMESPACE>
    priorityClassName: "build" (* see note below)
        user: <USER-NAME>
      - name: <JOB-NAME>
        image: <IMAGE-NAME>
      restartPolicy: Never
      schedulerName: runai-scheduler

To submit the job, run:

kubectl apply -f <FILE-NAME>


  • You can use either a regular Job or RunaiJob. The latter is a Run:ai object which solves various Kubernetes Bugs and provides a better naming for multiple pods in Hyper-Parameter Optimization scenarios
  • Using build in the priorityClassName field is equivalent to running a job via the CLI with a '--interactive' flag. To run a Train job, delete this line.
  • The runai submit CLI command includes many more flags. These flags can be correlated with Kubernetes API functions and added to the YAML above.

Using Fractional GPUs

To submit a Job with fractions of a GPU, replace <REQUESTED-GPUs> with a fraction in quotes. e.g.

limits: "0.5"

where "0.5" is the requested GPU fraction.

Mapping Additional Flags

Run:ai Command-Line runai submit has a significant number of flags. The easiest way to find out the mapping from a flag to the correct YAML attribute is to use the --dry-run flag.

For example, to find the location of the --large-shm flag, run:

> runai submit -i ubuntu --large-shm --dry-run
Template YAML file can be found at:

Delete Workloads

To delete a Run:ai workload, delete the Job:

kubectl delete runaijob <JOB-NAME>

See Also