feat: Add pipeline run parallelism config

[sduvvuri1603]

Summary

• Replace the previous semaphore/mutex knobs with a single pipeline_run_parallelism option on dsl.PipelineConfig. This lets the API server own the Argo semaphore lifecycles instead of expecting users to edit shared ConfigMaps—eliminating a Kubernetes-heavy workflow and ensuring keys align to <pipeline>/<version>.
• Thread the new field through SDK, compiler, and backend so the requested concurrency cap lands in Argo’s spec.parallelism.
• Add the pipeline_with_run_parallelism sample (three-item ParallelFor) to exercise the setting while leaving the workspace fixture focused on workspace behaviour.

Validation

• SDK and backend goldens now include the updated sample, showing consistent IR and Argo outputs with the parallelism limit.
• Built custom API server and driver images from this branch, loaded them into a kind cluster, ran the sample, and confirmed that the number of simultaneously running component pods never exceeded the configured limit.
• Added the parallelism validation helper to the e2e suite (e2e_utils.go + invocation in pipeline_e2e_test.go), rebuilt the test cluster with the fresh backend images, exercised the focused pipeline_run_parallelism scenario, and then ran the end-to-end suite to confirm the new check passes with the concurrency cap enforced.

Follow up to PR - remove unused semaphore_key and mutex_name fields

[google-oss-prow]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
Once this PR has been reviewed and has the lgtm label, please assign chensun for approval. For more information see the Kubernetes Code Review Process.

The full list of commands accepted by this bot can be found here.

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[google-oss-prow]

Hi @sduvvuri1603. Thanks for your PR.

I'm waiting for a kubeflow member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[alyssacgoins]

/retest

[google-oss-prow]

@sduvvuri1603: Cannot trigger testing until a trusted user reviews the PR and leaves an /ok-to-test message.

In response to this:

/retest

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[hbelmiro]

/ok-to-test

[hbelmiro]

/retest

[VaniHaripriya]

It may be more appropriate to add this entry to the CHANGELOG.

[sduvvuri1603]

Sure, but will this be a part of a new section called "Unreleased Features" ? because I only see version release details in the file.

[VaniHaripriya]

I believe the PR will be included here as part of the release process. @mprahl , could you confirm if that’s correct?

[nsingla]

is this required given https://github.com/kubeflow/pipelines/pull/12442/files#diff-631f096829954a5226e87a37347ab04d3e551465fdf0371b37fd9f9363c0c156R120 will set it to None if value is None?

[sduvvuri1603]

Yes, it's right about the serialization part. But we need this guard to prevent a crash during initialization. Since init passes None by default, removing this check would cause it to hit the isinstance line and fail immediately. This just ensures we can safely create the object with no value set.

[nsingla]

I guess then you can just add 1 if statment:

        if value:
          if not isinstance(value, int):
              raise ValueError(
                  'pipeline_run_parallelism must be an integer if specified.')
          if value <= 0:
              raise ValueError(
                  'pipeline_run_parallelism must be a positive integer.')
       self._pipeline_run_parallelism = value

[sduvvuri1603]

Updated code with suggested change

[nsingla]

can we actually set this to None here if we have an explicit test to test +ve values?

[sduvvuri1603]

Yep, That makes sense. Will set this to none

[nsingla]

@sduvvuri1603 can you please add what this config is suppose to do, to the PR description? and a section about how you;ve validated the functionality.

[nsingla]

Isn't 7 too high when the number of tasks in this pipeline is just 1? May be you should add more components to it or add a parallelFor loop and iterate over > pipeline_run_parallelism constants, so that we can validate that the config actually works.
Also what validation logic did you add to confirm the number of tasks created for a pipeline with this config?

[sduvvuri1603]

This specific test case is part of the SDK compilation suite to verify that the pipeline_run_parallelism field is correctly serialized from the Python SDK into the compiled YAML's PlatformSpec. It relies on the 'Golden File' comparison for validation here (ensuring the YAML contains pipelineRunParallelism: 7 correctly populated)

(It is not related to actual runtime limit covered by the backend integration tests where we submit these workflows to Argo is my understanding) Pls Lmk if this is correct!

[nsingla]

Any pipeline yaml file in this directory will be part of the end to end tests, so yes, the workflow will get submitted to argo.

[sduvvuri1603]

Updated code with ParallelFor loop and reduced pipeline_run_paralellism to '2'

[nsingla]

/lgtm