Enable 0 workers replicaSpec for pytorchjob #6201
Open
+78
−18
Conversation
This file contains 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
|
Thank you for opening this pull request! 🙌 These tips will help get your PR across the finish line:
|
Code Review Agent Run #29781fActionable Suggestions - 1
Review Details
AI Code Review powered by |
Changelist by BitoThis pull request implements the following key changes.
|
flyte-bot
reviewed
Jan 28, 2025
flyteplugins/go/tasks/plugins/k8s/kfoperators/pytorch/pytorch.go
Outdated
Show resolved
Hide resolved
Code Review Agent Run #c188b6Actionable Suggestions - 0Review Details
AI Code Review powered by |
pingsutw
reviewed
Jan 29, 2025
| @@ -145,16 +145,17 @@ func (p pytorchOperatorResourceHandler) BuildResource(ctx context.Context, taskC | |||
| "Invalid TaskSpecification, unsupported task template version [%v] key", taskTemplate.GetTaskTypeVersion()) | |||
| } | |||
|
|
|||
| if *workerReplicaSpec.Replicas <= 0 { | |||
| return nil, fmt.Errorf("number of workers must be greater than 0") | |||
| jobSpec := kubeflowv1.PyTorchJobSpec{} | |||
There was a problem hiding this comment.
Could we add a small unit test?
There was a problem hiding this comment.
Hey @pingsutw, I've removed TestBuildResourcePytorchV1WithZeroWorker as it was intended to fail in the case corrected with this PR. Replaced it with the TestBuildResourcePytorchV1WithDifferentWorkersNumber test, thank you.
There was a problem hiding this comment.
One unit test is failing. otherwise, lgtm.
There was a problem hiding this comment.
Should be fixed now, thanks.
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #6201 +/- ##
==========================================
- Coverage 37.08% 37.07% -0.01%
==========================================
Files 1318 1318
Lines 132707 132711 +4
==========================================
+ Hits 49208 49209 +1
- Misses 79244 79247 +3
Partials 4255 4255
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. |
punkerpunker
force-pushed
the
flyteplugins/allow-zero-worker-pytorchjob
branch
from
a683ad3 to
33ab59b
Compare
punkerpunker
force-pushed
the
flyteplugins/allow-zero-worker-pytorchjob
branch
from
33ab59b to
7464db6
Compare
Code Review Agent Run #59b7b8Actionable Suggestions - 1
Review Details
AI Code Review powered by |
flyte-bot
reviewed
Jan 29, 2025
Comment on lines
+148
to
+151
| jobSpec := kubeflowv1.PyTorchJobSpec{} | ||
| replicaSpecs := map[commonOp.ReplicaType]*commonOp.ReplicaSpec{ | ||
| kubeflowv1.PyTorchJobReplicaTypeMaster: masterReplicaSpec, | ||
| } |
There was a problem hiding this comment.
Consider validating masterReplicaSpec before use
Consider validating that masterReplicaSpec is not nil before using it in the replicaSpecs map. A nil masterReplicaSpec could cause runtime issues.
Code suggestion
Check the AI-generated fix before applying
Suggested change
| jobSpec := kubeflowv1.PyTorchJobSpec{} | |
| replicaSpecs := map[commonOp.ReplicaType]*commonOp.ReplicaSpec{ | |
| kubeflowv1.PyTorchJobReplicaTypeMaster: masterReplicaSpec, | |
| } | |
| jobSpec := kubeflowv1.PyTorchJobSpec{} | |
| if masterReplicaSpec == nil { | |
| return nil, fmt.Errorf("master replica spec cannot be nil") | |
| } | |
| replicaSpecs := map[commonOp.ReplicaType]*commonOp.ReplicaSpec{ | |
| kubeflowv1.PyTorchJobReplicaTypeMaster: masterReplicaSpec, | |
| } |
Code Review Run #59b7b8
Is this a valid issue, or was it incorrectly flagged by the Agent?
- it was incorrectly flagged
pingsutw
previously approved these changes
Jan 29, 2025
Signed-off-by: punker <[email protected]>
Co-authored-by: Flyte Bot <[email protected]> Signed-off-by: Gleb <[email protected]> Signed-off-by: punker <[email protected]>
Signed-off-by: punker <[email protected]>
Signed-off-by: punker <[email protected]>
Signed-off-by: punker <[email protected]>
punkerpunker
force-pushed
the
flyteplugins/allow-zero-worker-pytorchjob
branch
from
a71e27b to
644ce84
Compare
Code Review Agent Run #216032Actionable Suggestions - 0Review Details
AI Code Review powered by |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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 the changes needed?
Currently, PyTorchJob allows not to specify Worker in case only one worker (Master) is needed for the job. Current functionality of the plugin doesn't allow to do so, however, in some cases we still want to achieve a Master-only PyTorchJob.
What changes were proposed in this pull request?
Made a change so
pytorchplugin is no longer complains about a 0-replica worker, but instead - if replicas=0 is passed, plugin just doesn't render it.I understand this isn't a best design for the moment, but that is a least-invasive approach to achieve it without changing e.g
pyflytevalidation behavior (if e.g we accept that worker=None could be passed to the PyTorchtask_config.How was this patch tested?
We are running the fork code in our production environment
Check all the applicable boxes
Summary by Bito
Enhanced PyTorch operator plugin to support master-only configurations by removing the requirement for worker replicas. Modified replica specification logic to make worker specs optional. Updated test suite to validate both zero-worker and single-worker configurations.Unit tests added: True
Estimated effort to review (1-5, lower is better): 2