Add --node-selector Flag for shp build/buildrun Commands

[rxinui]

Changes

Fixes #260

Add CLI flag --node to set .spec.nodeSelector to a Build or BuildRun

❗ IMPORTANT: depends on PR 304

PR's codebase is tracking PR 304 by @karanibm6 as .spec.nodeSelector is only defined on builds.shipwright.io/v1beta1 as a consequence of the current main branch for CLI relying on builds.shipwright.io/v1alpha1 (see PR 1683)

Submitter Checklist

• Includes tests if functionality changed/was added
• Includes docs if changes are user-facing
• Set a kind label on this PR
• Release notes block has been filled in, or marked NONE

See the contributor guide
for details on coding conventions, github and prow interactions, and the code review process.

Release Notes

Add flag --node-selector to set the `.spec.nodeSelector` for `build` and `buildrun` on create

[rxinui]

cc: @adambkaplan

sorry for the delay

[adambkaplan]

Great work so far!

I added my review of the items that appear directly related to the node selector change. The bulk of my comments are related to changing the name of the flag to --node-selector. Otherwise this is in very good shape. Once the v1beta1 API PR is merged this can be rebased + unblocked.

[adambkaplan]

/ok-to-test

[rxinui]

As requested, a new bats test is added for shp build run <mybuild> --node-selector which works as expected. However, there is a use case where I have doubts.

Context: since --node-selector is added to shp build create and shp build run, the field .spec.nodeSelector is can now be defined on Build and BuildRun.

Doubt: when running successively the following

# create mybuild with pre-defined nodeSelector
shp build create mybuild --node-selector="foo=bar" --source-git-url=https://github.com/shipwright-io/sample-go --output-image=my-fake-image
# run mybuild
shp build run mybuild

Outcome: BuildRun does not have .spec.nodeSelector specified

apiVersion: shipwright.io/v1beta1
kind: BuildRun
metadata:
  creationTimestamp: "2025-04-23T14:40:59Z"
  generateName: mybuild-
  generation: 1
  labels:
    build.shipwright.io/generation: "1"
    build.shipwright.io/name: mybuild
  name: mybuild-4blqb
  namespace: default
  resourceVersion: "510650"
  uid: 2d74cd53-88fd-469c-a44a-1a5d744027f5
spec: # missing .nodeSelector
  build:
    name: mybuild
status:
  buildSpec:
    nodeSelector: # present in status
      foo: bar
    output:
      image: my-fake-image
# ...

Expectation:

• A. It is the expected behaviour, where pre-defined .spec.nodeSelector should not be embedded within BuildRun unless specified on shp build run --node-selector?
• B. It is a missing feature, we must include .spec.nodeSelector pre-defined by shp build create command within the BuildRun unless it is overriden by shp build run --node-selector (current behaviour)

In case of A, we should consider allowing --node-selector only on shp build run and shp buildrun create.

[adambkaplan]

Expectation:

• A. It is the expected behaviour, where pre-defined .spec.nodeSelector should not be embedded within BuildRun unless specified on shp build run --node-selector?
• B. It is a missing feature, we must include .spec.nodeSelector pre-defined by shp build create command within the BuildRun unless it is overriden by shp build run --node-selector (current behaviour)

In case of A, we should consider allowing --node-selector only on shp build run and shp buildrun create.

This is expected behavior. Our API has the following convention for settings that are "shared" across Build and BuildRun:

1. The value in the BuildRun is chosen first.
2. If no value exists in the BuildRun, but it references a Build object, use the value in the referenced Build object.
3. If no value exists in the referenced Build object, use the "default" behavior for a nil/empty value.

The status field is used to communicate what the controller inferred from the BuildRun and Build settings.

[adambkaplan]

In case of A, we should consider allowing --node-selector only on shp build run and shp buildrun create

shp build run is in many ways an alias to shp buildrun create. Not 100% identical, but they both result in a workload being executed. Sharing these flags ensures feature parity.

[rxinui]

In case of A, we should consider allowing --node-selector only on shp build run and shp buildrun create

shp build run is in many ways an alias to shp buildrun create. Not 100% identical, but they both result in a workload being executed. Sharing these flags ensures feature parity.

Thank you for the clarification @adambkaplan

Everything is sorted on my side. Waiting for #304 to be merged to proceed further.

[rxinui]

@adambkaplan @SaschaSchwarze0 I have rebased to the latest commit which includes v1beta1. Please trigger the /ok-to-test . Note: this blocks PR 311

The checks are green on my forked repository.

[adambkaplan]

/approve

Code changes itself look good! There are only two items IMO that need to be fixed before this merges:

1. Revert the changes in the vendor directory (unclear where this newline change came from).
2. Squash commits.

[openshift-ci]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: adambkaplan

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

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [adambkaplan]

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

[rxinui]

@adambkaplan ready to be merged, commit is squashed.

[adambkaplan]

/lgtm

Looks good!