[Autoscaler][V2] Add top-level Resources and Labels field to KubeRay Autoscaling config #57260
+229
−42
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: Ryan O'Leary <[email protected]>
Signed-off-by: Ryan O'Leary <[email protected]>
Contributor
Author
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
This comment was marked as outdated.
This comment was marked as outdated.
Sorry, something went wrong.
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Code Review
This pull request adds support for top-level Resources and Labels fields in the KubeRay autoscaling configuration, which is a great enhancement for usability and alignment with upcoming KubeRay features. The implementation correctly prioritizes these new fields over older configuration methods. The accompanying tests are thorough and cover the new functionality well. I've identified a few areas for improvement, mainly concerning type consistency in function signatures and return values, which could prevent potential bugs and improve code clarity. There is also a minor documentation error in a docstring. Overall, this is a solid contribution.
ray-gardener
bot
added
kuberay
4 tasks
MengjinYan
reviewed
Oct 8, 2025
Signed-off-by: Ryan O'Leary <[email protected]>
This comment was marked as outdated.
This comment was marked as outdated.
Sorry, something went wrong.
Signed-off-by: Ryan O'Leary <[email protected]>
This comment was marked as outdated.
This comment was marked as outdated.
Sorry, something went wrong.
Signed-off-by: Ryan O'Leary <[email protected]>
35 tasks
MengjinYan
reviewed
Oct 15, 2025
Contributor
There was a problem hiding this comment.
Just one curious question. Overall looks good!
Also you might need to fix the doc lint issue:
[2025-10-11T01:27:22Z] ./python/ray/autoscaler/_private/kuberay/autoscaling_config.py: f"Ignoring labels: {labels_str} set in rayStartParams. Group labels are supported in the top-level Labels field starting in Kuberay v1.5"
[2025-10-11T01:27:22Z] ./python/ray/tests/kuberay/test_autoscaling_config.py: "Ignoring labels: ray.io/accelerator-type=TPU-V4 set in rayStartParams. Group labels are supported in the top-level Labels field starting in Kuberay v1.5",
[2025-10-11T01:27:22Z] ./python/ray/tests/kuberay/test_autoscaling_config.py: "Ignoring labels: instance-type=n2 set in rayStartParams. Group labels are supported in the top-level Labels field starting in Kuberay v1.5",
[2025-10-11T01:27:22Z] ******************************
[2025-10-11T01:27:22Z] *** Misspelled word found! ***
[2025-10-11T01:27:22Z] ******************************
[2025-10-11T01:27:22Z] Please fix the capitalization/spelling of "Kuberay" in the above files.```
Signed-off-by: Ryan O'Leary <[email protected]>
Contributor
Author
Oh I think I've seen this before, it's because I failed to capitalize the "R" in "KubeRay". Should be fixed with d0ea801. |
Contributor
|
@rueian if you want to take a look from autoscaler's perspective. |
Contributor
justinyeh1995
pushed a commit
to justinyeh1995/ray
that referenced
this pull request
Oct 20, 2025
…Autoscaling config (ray-project#57260) In KubeRay v1.5, new structured `Resources` and `Labels` fields will be added to both the `HeadGroupSpec` and `WorkerGroupSpec` to enable users to explicitly define these values without relying on `rayStartParams`. This support is implemented in ray-project/kuberay#4106. In order for this change to work with the Ray autoscaler, this PR adds support for checking the `resources` and `labels` field (with precedence over the values `rayStartParams` and k8s constainer spec) and generating the resulting autoscaling config. This change is compatible with older versions of KubeRay that do not have these new fields, since we fall back to the previous logic when the top-level fields aren't specified. Example: For a group spec specified like this: ``` workerGroupSpecs: - groupName: worker-group-1 replicas: 1 resources: GPU: 8 CPU: 16 labels: ray.io/zone: us-west2-a ray.io/region: us-west2 ``` The resulting autoscaling config for worker-group-1 would be: ``` "worker-group-1": { "labels": {" ray.io/zone": "us-west2-a", " ray.io/region": "us-west2"}, ... "resources": { "CPU": 16, GPU: 8, }, }, ``` ## Related issue number Contributes to ray-project#51564 --------- Signed-off-by: Ryan O'Leary <[email protected]>
xinyuangui2
pushed a commit
to xinyuangui2/ray
that referenced
this pull request
Oct 22, 2025
…Autoscaling config (ray-project#57260) In KubeRay v1.5, new structured `Resources` and `Labels` fields will be added to both the `HeadGroupSpec` and `WorkerGroupSpec` to enable users to explicitly define these values without relying on `rayStartParams`. This support is implemented in ray-project/kuberay#4106. In order for this change to work with the Ray autoscaler, this PR adds support for checking the `resources` and `labels` field (with precedence over the values `rayStartParams` and k8s constainer spec) and generating the resulting autoscaling config. This change is compatible with older versions of KubeRay that do not have these new fields, since we fall back to the previous logic when the top-level fields aren't specified. Example: For a group spec specified like this: ``` workerGroupSpecs: - groupName: worker-group-1 replicas: 1 resources: GPU: 8 CPU: 16 labels: ray.io/zone: us-west2-a ray.io/region: us-west2 ``` The resulting autoscaling config for worker-group-1 would be: ``` "worker-group-1": { "labels": {" ray.io/zone": "us-west2-a", " ray.io/region": "us-west2"}, ... "resources": { "CPU": 16, GPU: 8, }, }, ``` ## Related issue number Contributes to ray-project#51564 --------- Signed-off-by: Ryan O'Leary <[email protected]> Signed-off-by: xgui <[email protected]>
elliot-barn
pushed a commit
that referenced
this pull request
Oct 23, 2025
…Autoscaling config (#57260) In KubeRay v1.5, new structured `Resources` and `Labels` fields will be added to both the `HeadGroupSpec` and `WorkerGroupSpec` to enable users to explicitly define these values without relying on `rayStartParams`. This support is implemented in ray-project/kuberay#4106. In order for this change to work with the Ray autoscaler, this PR adds support for checking the `resources` and `labels` field (with precedence over the values `rayStartParams` and k8s constainer spec) and generating the resulting autoscaling config. This change is compatible with older versions of KubeRay that do not have these new fields, since we fall back to the previous logic when the top-level fields aren't specified. Example: For a group spec specified like this: ``` workerGroupSpecs: - groupName: worker-group-1 replicas: 1 resources: GPU: 8 CPU: 16 labels: ray.io/zone: us-west2-a ray.io/region: us-west2 ``` The resulting autoscaling config for worker-group-1 would be: ``` "worker-group-1": { "labels": {" ray.io/zone": "us-west2-a", " ray.io/region": "us-west2"}, ... "resources": { "CPU": 16, GPU: 8, }, }, ``` ## Related issue number Contributes to #51564 --------- Signed-off-by: Ryan O'Leary <[email protected]> Signed-off-by: elliot-barn <[email protected]>
landscapepainter
pushed a commit
to landscapepainter/ray
that referenced
this pull request
Nov 17, 2025
…Autoscaling config (ray-project#57260) In KubeRay v1.5, new structured `Resources` and `Labels` fields will be added to both the `HeadGroupSpec` and `WorkerGroupSpec` to enable users to explicitly define these values without relying on `rayStartParams`. This support is implemented in ray-project/kuberay#4106. In order for this change to work with the Ray autoscaler, this PR adds support for checking the `resources` and `labels` field (with precedence over the values `rayStartParams` and k8s constainer spec) and generating the resulting autoscaling config. This change is compatible with older versions of KubeRay that do not have these new fields, since we fall back to the previous logic when the top-level fields aren't specified. Example: For a group spec specified like this: ``` workerGroupSpecs: - groupName: worker-group-1 replicas: 1 resources: GPU: 8 CPU: 16 labels: ray.io/zone: us-west2-a ray.io/region: us-west2 ``` The resulting autoscaling config for worker-group-1 would be: ``` "worker-group-1": { "labels": {" ray.io/zone": "us-west2-a", " ray.io/region": "us-west2"}, ... "resources": { "CPU": 16, GPU: 8, }, }, ``` ## Related issue number Contributes to ray-project#51564 --------- Signed-off-by: Ryan O'Leary <[email protected]>
Aydin-ab
pushed a commit
to Aydin-ab/ray-aydin
that referenced
this pull request
Nov 19, 2025
…Autoscaling config (ray-project#57260) In KubeRay v1.5, new structured `Resources` and `Labels` fields will be added to both the `HeadGroupSpec` and `WorkerGroupSpec` to enable users to explicitly define these values without relying on `rayStartParams`. This support is implemented in ray-project/kuberay#4106. In order for this change to work with the Ray autoscaler, this PR adds support for checking the `resources` and `labels` field (with precedence over the values `rayStartParams` and k8s constainer spec) and generating the resulting autoscaling config. This change is compatible with older versions of KubeRay that do not have these new fields, since we fall back to the previous logic when the top-level fields aren't specified. Example: For a group spec specified like this: ``` workerGroupSpecs: - groupName: worker-group-1 replicas: 1 resources: GPU: 8 CPU: 16 labels: ray.io/zone: us-west2-a ray.io/region: us-west2 ``` The resulting autoscaling config for worker-group-1 would be: ``` "worker-group-1": { "labels": {" ray.io/zone": "us-west2-a", " ray.io/region": "us-west2"}, ... "resources": { "CPU": 16, GPU: 8, }, }, ``` ## Related issue number Contributes to ray-project#51564 --------- Signed-off-by: Ryan O'Leary <[email protected]> Signed-off-by: Aydin Abiar <[email protected]>
Future-Outlier
pushed a commit
to Future-Outlier/ray
that referenced
this pull request
Dec 7, 2025
…Autoscaling config (ray-project#57260) In KubeRay v1.5, new structured `Resources` and `Labels` fields will be added to both the `HeadGroupSpec` and `WorkerGroupSpec` to enable users to explicitly define these values without relying on `rayStartParams`. This support is implemented in ray-project/kuberay#4106. In order for this change to work with the Ray autoscaler, this PR adds support for checking the `resources` and `labels` field (with precedence over the values `rayStartParams` and k8s constainer spec) and generating the resulting autoscaling config. This change is compatible with older versions of KubeRay that do not have these new fields, since we fall back to the previous logic when the top-level fields aren't specified. Example: For a group spec specified like this: ``` workerGroupSpecs: - groupName: worker-group-1 replicas: 1 resources: GPU: 8 CPU: 16 labels: ray.io/zone: us-west2-a ray.io/region: us-west2 ``` The resulting autoscaling config for worker-group-1 would be: ``` "worker-group-1": { "labels": {" ray.io/zone": "us-west2-a", " ray.io/region": "us-west2"}, ... "resources": { "CPU": 16, GPU: 8, }, }, ``` ## Related issue number Contributes to ray-project#51564 --------- Signed-off-by: Ryan O'Leary <[email protected]> Signed-off-by: Future-Outlier <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Why are these changes needed?
In KubeRay v1.5, new structured
ResourcesandLabelsfields will be added to both theHeadGroupSpecandWorkerGroupSpecto enable users to explicitly define these values without relying onrayStartParams. This support is implemented in ray-project/kuberay#4106. In order for this change to work with the Ray autoscaler, this PR adds support for checking theresourcesandlabelsfield (with precedence over the valuesrayStartParamsand k8s constainer spec) and generating the resulting autoscaling config. This change is compatible with older versions of KubeRay that do not have these new fields, since we fall back to the previous logic when the top-level fields aren't specified.Example:
For a group spec specified like this:
The resulting autoscaling config for worker-group-1 would be:
Related issue number
Contributes to #51564
Checks
git commit -s) in this PR.method in Tune, I've added it in
doc/source/tune/api/under thecorresponding
.rstfile.