[None][fix] drop the placeholder MicroBatchScheduler that shadows the ABC

[Anai-Guo]

Description

tensorrt_llm/_torch/pyexecutor/scheduler/scheduler.py binds the module-level name
MicroBatchScheduler twice:

# line 474 - the real contract
class MicroBatchScheduler(ABC):
    @abstractmethod
    def schedule(self, active_requests, inflight_request_ids): ...
        # to be aligned with MicroBatchScheduler::scheduleRequests
        # in cpp/tensorrt_llm/batch_manager/microBatchScheduler.h

# line 867 - placeholder
class MicroBatchScheduler:
    """Base class to match structure."""

Python executes both class statements, so the second one silently wins. The
placeholder came in with dbb858a ([TRTLLM-10029][scheduler] Re-implement MicroBatchScheduler and CapacityScheduler…) alongside PyMicroBatchScheduler, which
is declared right after it and therefore inherits the placeholder rather than the ABC.

This matters because MicroBatchScheduler is a public export — it is listed in
tensorrt_llm/_torch/pyexecutor/scheduler/__init__.py's __all__, so
from tensorrt_llm._torch.pyexecutor.scheduler import MicroBatchScheduler hands the
caller the empty placeholder.

Effect

Lifting the three class statements out of the file in their upstream order (headers

• method signatures only, ABC/abstractmethod bound for real, no torch needed):

BEFORE (upstream main)
  MicroBatchScheduler is an ABC ............ False
  MicroBatchScheduler.schedule exists ...... False
  issubclass(BindMicroBatchScheduler, MBS) . False
  issubclass(PyMicroBatchScheduler,   MBS) . True
  PyMicroBatchScheduler.__mro__ ............ ['PyMicroBatchScheduler', 'MicroBatchScheduler', 'object']

AFTER  (placeholder deleted)
  MicroBatchScheduler is an ABC ............ True
  MicroBatchScheduler.schedule exists ...... True
  issubclass(BindMicroBatchScheduler, MBS) . True
  issubclass(PyMicroBatchScheduler,   MBS) . True
  PyMicroBatchScheduler.__mro__ ............ ['PyMicroBatchScheduler', 'MicroBatchScheduler', 'ABC', 'object']

So today BindMicroBatchScheduler — the binding used by _util.py and
auto_deploy/shim/ad_executor.py — is not a subclass of the exported
MicroBatchScheduler, and neither scheduler implementation gets the
@abstractmethod enforcement the ABC was written to provide.

Fix

Delete the two-line placeholder (4 lines including the surrounding blank lines,
+0 / -4, one file). PyMicroBatchScheduler already implements schedule(), so it
satisfies the ABC as-is and nothing else changes: no method bodies, no signatures, no
call sites.

Test Coverage

No behaviour change, so no new test. Existing tests/unittest/_torch/executor/test_py_scheduler.py
covers PyMicroBatchScheduler and continues to apply unchanged.

PR Checklist

• PR title is in the format [JIRA ticket/NVBugs ID/GitHub issue/None][type] Summary
• Commit is signed off (DCO)
• Change is limited to the described fix

---

AI-assisted (found by a duplicate-module-level-class sweep; the effect table above
was produced by lifting the real class statements out of the upstream file, and I
reviewed the change before submitting).

🤖 Generated with Claude Code

Dev Engineer Review

Removed the duplicate MicroBatchScheduler placeholder. The original abstract base class remains exported, so schedule() enforcement and BindMicroBatchScheduler subclass recognition are restored. No other source behavior or API changes were identified.

QA Engineer Review

No test changes.

Per-File QA Perspective

• tensorrt_llm/_torch/pyexecutor/scheduler/scheduler.py: Verify that MicroBatchScheduler exposes the abstract schedule() contract and that BindMicroBatchScheduler is recognized as its subclass. Existing scheduler tests should remain applicable.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 419a092a-ed09-4d96-991c-30a693c8a016

📥 Commits

Reviewing files that changed from the base of the PR and between a85c26c and 774625a.

📒 Files selected for processing (1)

• tensorrt_llm/_torch/pyexecutor/scheduler/scheduler.py

💤 Files with no reviewable changes (1)

• tensorrt_llm/_torch/pyexecutor/scheduler/scheduler.py

Included review availability: Your plan provides up to 12 included reviews per hour; 11 remain after this review.

---

Walkthrough

The change removes an empty MicroBatchScheduler placeholder from the scheduler module. The functional scheduler abstraction remains.

Changes

Scheduler cleanup

Layer / File(s)	Summary
Remove obsolete scheduler declaration tensorrt_llm/_torch/pyexecutor/scheduler/scheduler.py	Removes the empty MicroBatchScheduler placeholder. The functional scheduler abstraction remains.

Priority: ⬇️ Low

Estimated code review effort: 1 (Trivial) | ~2 minutes

Change: Bug fix

Suggested reviewers: bowenfu

Merge Risk: ⚪ Minimal · up to 77462

This cleanup restores the intended scheduler abstraction without introducing an identified merge risk.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title follows the required [ticket][type] format and clearly states that the placeholder MicroBatchScheduler is removed to prevent it from shadowing the ABC.
Description check	✅ Passed	The description explains the issue, root cause, fix, impact, and test coverage. It includes the required Description and Test Coverage sections and documents the relevant checklist items.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check. Docstring coverage is scoped to functions touched by this diff. Analyzed 0 functions across 0…
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[mikeiovine]

You have many small PRs open; please consolidate