Creating and working with run groups - AWS HealthOmics

Creating and working with run groups

You can optionally create a run group to limit the compute resources for the runs that you add to the group. Run groups can help you:

  • Queue your runs so that you don’t exceed service limits.

  • Catch run-away tasks by setting a maximum run duration limit.

  • Manage the priority of each run so that the most important runs complete first

If you set the maximum concurrent vCPU, GPU, or runs, run tasks will queue when the limit is reached. If you set a maximum run duration, the run fails if it exceeds the maximum duration.

Use the run priority setting to establish priority within a run group.

Service limits take precedence over run group limits. For instance, if a run group limit is set higher than your service limit, the service limit will apply first.

Run priority

You can use run priority to establish the priority of runs in a run group.

If multiple runs have the same priority, the run that started first has the higher priority.

You can also set a priority for a run that isn't in a run group. The priority is compared with the priorities of all other runs that aren't in a run group

You set run priority when you start the workflow run. For more information, see Starting a workflow run.

Create a run group (console)

To create a run group
  1. Open the HealthOmics console

  2. In the left navigation pane, choose Run groups.

  3. On the Run groups page, choose Create run group.

  4. On the Create run group details page, provide the following information

    • Run group name - A unique name for this run group.

    • Max vCPU for concurrent runs - The maximum number of vCPUs that can run concurrently across all active runs in the run group.

    • Max GPUs - The maximum number of GPUs that can run concurrently across all active runs in the run group.

    • Max run time (mins) per run - The maximum time for each run (in minutes). If a run exceeds the maximum run time, the run fails automatically.

    • Max concurrent runs - The maximum number of runs that can be running at the same time.

  5. (optional) You can add up to 50 tags to the run group.

  6. Choose Create run group.

Create a run group (API)

To create a run group, use the create-run-group API operation to create a run group named TestRunGroup. The following example sets a maximum of 20 CPUs, 10 GPUs, 5 runs, and a maximum run duration of 600 minutes.

aws omics create-run-group --name TestRunGroup \ --max-cpus 20 \ --max-gpus 10 \ --max-duration 600 \ --max-runs 5

The response from this API operation includes the ID of the newly created RunGroup.

{ "arn": "arn:aws:omics:us-west-2:12345678901:runGroup/2839621", "id": "2839621", "tags": {} }

To get additional information about the run group, use this ID with the get-run-group API operation, as shown in the following example.

aws omics get-run-group --id run group id

The response includes the limit settings for the run group and the assigned tags.

{ "arn": "arn:aws:omics:us-west-2:776893852117:runGroup/2839621", "id": "2839621", "name": "TestRunGroup", "maxCpus": 20, "maxRuns": 5, "maxDuration": 600, "creationTime": "2024-06-12T15:35:39.191730+00:00", "tags": {}, "maxGpus": 10 }

You can also use the list-run-group API operation to view all created run groups.

aws omics list-run-groups