feat(sdk/backend): Support forwarding task configuration to external workloads #12185
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
|
Skipping CI for Draft Pull Request. |
google-oss-prow
bot
requested review from
DharmitD,
connor-mccarthy,
droctothorpe and
zazulam
mprahl
force-pushed
the
trainer-integration
branch
4 times, most recently
from
3313faa to
a56e3a8
Compare
mprahl
commented
Aug 27, 2025
mprahl
commented
Aug 27, 2025
mprahl
commented
Aug 27, 2025
mprahl
force-pushed
the
trainer-integration
branch
4 times, most recently
from
8f56491 to
8daf9a9
Compare
|
/cc @HumairAK |
mprahl
changed the title
HumairAK
removed request for
DharmitD,
connor-mccarthy,
gmfrasca and
rimolive
HumairAK
reviewed
Sep 2, 2025
mprahl
force-pushed
the
trainer-integration
branch
9 times, most recently
from
d7b4db5 to
a56f03f
Compare
Signed-off-by: mprahl <[email protected]>
mprahl
force-pushed
the
trainer-integration
branch
2 times, most recently
from
5fee341 to
ed52a40
Compare
HumairAK
requested changes
Sep 4, 2025
api/v2alpha1/pipeline_spec.proto
Outdated
| // Throwaway default value. | ||
| NONE = 0; | ||
| // Indicates that the resource limits and requests should be passed through to the external workload. | ||
| // Be cautious about also setting apply_to_task=true since that may double apply the |
There was a problem hiding this comment.
Suggested change
| // Be cautious about also setting apply_to_task=true since that may double apply the | |
| // Be cautious about also setting apply_to_task=true since that will double apply | |
| pip_trusted_hosts: Optional[List[str]] = None, | ||
| use_venv: bool = False, | ||
| additional_funcs: Optional[List[Callable]] = None, | ||
| task_config_passthroughs: Optional[List[Union[TaskConfigPassthrough, |
There was a problem hiding this comment.
missing this parameter in the docstring
backend/src/v2/driver/driver_test.go
Outdated
Comment on lines
680
to
683
| CpuLimit: 2.0, | ||
| MemoryLimit: 1.5, | ||
| CpuRequest: 1.0, | ||
| MemoryRequest: 0.65, |
There was a problem hiding this comment.
can we use the non-deprecated fields?
Introduce dsl.TaskConfig to capture task-level resource settings for pass-through to externally launched workloads (e.g., trainers). The driver now updates container execution and Kubernetes handling to inject the Kubernetes config into executor inputs without applying disallowed fields to the task pod. Signed-off-by: mprahl <[email protected]>
Signed-off-by: mprahl <[email protected]>
mprahl
force-pushed
the
trainer-integration
branch
from
ed52a40 to
5393888
Compare
|
/lgtm niceee |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: HumairAK 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:
Approvers can indicate their approval by writing |
9 tasks
krishanbhasin-px
pushed a commit
to krishanbhasin-px/kubeflow-pipelines
that referenced
this pull request
Sep 18, 2025
…workloads (kubeflow#12185) * Add missing PULL_NUMBER env var to sample tests CI Signed-off-by: mprahl <[email protected]> * Use the local API package for the upgrade test Signed-off-by: mprahl <[email protected]> * Support forwarding task configuration to external workloads Introduce dsl.TaskConfig to capture task-level resource settings for pass-through to externally launched workloads (e.g., trainers). The driver now updates container execution and Kubernetes handling to inject the Kubernetes config into executor inputs without applying disallowed fields to the task pod. Signed-off-by: mprahl <[email protected]> * Fix dependencies issue in Read The Docs build Signed-off-by: mprahl <[email protected]> --------- Signed-off-by: mprahl <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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.
Description of your changes:
This PR introduces a powerful new feature that allows KFP tasks to forward their task configuration (resources, environment variables, node selectors, tolerations, affinity, volumes) to external workloads like Kubeflow TrainJobs, rather than applying them only to the task pod itself. This is particularly valuable for scenarios where the KFP task is launching another Kubernetes workload (e.g.
TrainJob,Job, etc).Checklist: