sigs.k8s.io/kueue@v0.6.2/site/content/en/docs/tasks/run_kubeflow_jobs/run_mpijobs.md (about)

     1  ---
     2  title: "Run a MPIJob"
     3  date: 2023-05-16
     4  weight: 6
     5  description: >
     6    Run a Kueue scheduled MPIJob
     7  ---
     8  
     9  This page shows how to leverage Kueue's scheduling and resource management capabilities when running [MPI Operator](https://www.kubeflow.org/docs/components/training/mpi/) MPIJobs.
    10  
    11  This guide is for [batch users](/docs/tasks#batch-user) that have a basic understanding of Kueue. For more information, see [Kueue's overview](/docs/overview).
    12  
    13  ## Before you begin
    14  
    15  Check [administer cluster quotas](/docs/tasks/administer_cluster_quotas) for details on the initial cluster setup.
    16  
    17  Check [the MPI Operator installation guide](https://github.com/kubeflow/mpi-operator#installation).
    18  
    19  You can [modify kueue configurations from installed releases](/docs/installation#install-a-custom-configured-released-version) to include MPIJobs as an allowed workload.  
    20  
    21  ## MPI Operator definition
    22  
    23  ### a. Queue selection
    24  
    25  The target [local queue](/docs/concepts/local_queue) should be specified in the `metadata.labels` section of the MPIJob configuration.
    26  
    27  ```yaml
    28  metadata:
    29    labels:
    30      kueue.x-k8s.io/queue-name: user-queue
    31  ```
    32  
    33  ### b. Optionally set Suspend field in MPIJobs
    34  
    35  ```yaml
    36  spec:
    37    runPolicy:
    38      suspend: true
    39  ```
    40  
    41  By default, Kueue will set `suspend` to true via webhook and unsuspend it when the MPIJob is admitted.
    42  
    43  ## Sample MPI Job
    44  
    45  This example is based on https://github.com/kubeflow/mpi-operator/blob/ccf2756f749336d652fa6b10a732e241a40c7aa6/examples/v2beta1/pi/pi.yaml.
    46  
    47  {{< include "examples/jobs/sample-mpijob.yaml" "yaml" >}}
    48  
    49  For equivalent instructions for doing this in Python, see [Run Python Jobs](/docs/tasks/run_python_jobs/#mpi-operator-job).