schedulers parameterized #1022
Merged
schedulers parameterized #1022
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
facebook-github-bot
added
the
CLA Signed
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D71023681 |
bobyangyf
force-pushed
the
export-D71023681
branch
from
76ef188 to
fce9143
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D71023681 |
kiukchung
reviewed
Mar 21, 2025
Contributor
kiukchung
left a comment
kiukchung
left a comment
There was a problem hiding this comment.
I don't fully understand this change. It seems like you want torchx.schedulers.api.Scheduler interface to be able to deal with both AppDef and PipelineDef in the same API. For instance:
def schedule(desc: Union[AppDef, PipelineDef], ...)
If this is the case, I would discourage you from overloading the scheduler APIs in such a way.
I'd have to fully understand what the intent here is before recommending a good path forward. Happy to meet and discuss.
lgarg26
approved these changes
Mar 23, 2025
Contributor
lgarg26
left a comment
lgarg26
left a comment
There was a problem hiding this comment.
Accepting to unblock ongoing intergration.
Summary: Parameterize Scheduler This not only more precisely expresses it's type (previously was unspecified generic variant of AppDryRunInfo[T] where T was unspecified; not it can be fully specified); This also allows us to inject PipelineDryRunInfo for schedulers that are pipeline schedulers Reviewed By: lgarg26 Differential Revision: D71023681
bobyangyf
force-pushed
the
export-D71023681
branch
from
fce9143 to
83bfe40
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D71023681 |
facebook-github-bot
merged commit 9814e75
into
meta-pytorch:main
18 of 24 checks passed
facebook-github-bot
pushed a commit
that referenced
this pull request
Nov 26, 2025
Summary: The `torchx.schedulers.Scheduler` interface was made overly generic to overload the concept of an "orchestration" scheduler in #1022. DAG orchestration has never been something torchx should've supported natively. Instead torchx AppDefs can be adapted to an orchestrator framework's notion of a "NodeDefinition". This way users can use job definitions defined using torchx's AppDef as a node in a DAG to string together applications in a specific execution order. This PR/Diff reverts the over-genericized `torchx.schedulers.Scheduler` back to supporting `AppDef` and `AppDryRunInfo` as the primitives. Differential Revision: D87901987
facebook-github-bot
pushed a commit
that referenced
this pull request
Nov 26, 2025
Summary: The `torchx.schedulers.Scheduler` interface was made overly generic to overload the concept of an "orchestration" scheduler in #1022. DAG orchestration has never been something torchx should've supported natively. Instead torchx AppDefs can be adapted to an orchestrator framework's notion of a "NodeDefinition". This way users can use job definitions defined using torchx's AppDef as a node in a DAG to string together applications in a specific execution order. This PR/Diff reverts the over-genericized `torchx.schedulers.Scheduler` back to supporting `AppDef` and `AppDryRunInfo` as the primitives. Differential Revision: D87901987
facebook-github-bot
pushed a commit
that referenced
this pull request
Nov 26, 2025
Summary: The `torchx.schedulers.Scheduler` interface was made overly generic to overload the concept of an "orchestration" scheduler in #1022. DAG orchestration has never been something torchx should've supported natively. Instead torchx AppDefs can be adapted to an orchestrator framework's notion of a "NodeDefinition". This way users can use job definitions defined using torchx's AppDef as a node in a DAG to string together applications in a specific execution order. This PR/Diff reverts the over-genericized `torchx.schedulers.Scheduler` back to supporting `AppDef` and `AppDryRunInfo` as the primitives. Reviewed By: AbishekS Differential Revision: D87901987
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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.
Summary:
Parameterize Scheduler
This not only more precisely expresses it's type (previously was unspecified generic variant of AppDryRunInfo[T] where T was unspecified; not it can be fully specified); This also allows us to inject PipelineDryRunInfo for schedulers that are pipeline schedulers
Differential Revision: D71023681