[None][fix] Revert 'Add TestServePrefixAwareScheduling base on LMBenchmark/synthetic-multi-round-qa'

[tburt-nv]

The kv_cache/test_prefix_aware_scheduling.py tests added by #13243 fail in CI.

Summary by CodeRabbit

• Refactor

  • Simplified internal prefix-reuse caching logic in the scheduler to improve efficiency.
  • Optimized node deletion behavior in data structure handling.
  • Updated block accounting APIs for more streamlined resource management.

• Tests

  • Removed integration and unit tests related to prefix-aware scheduling validation.
  • Updated test configurations to reflect changed testing scope.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: fffee15b-d8ed-48d8-a3cf-13aea8bd7122

📥 Commits

Reviewing files that changed from the base of the PR and between c30d9c7 and bf41aa5.

📒 Files selected for processing (7)

• cpp/include/tensorrt_llm/batch_manager/templatedTrie.h
• tensorrt_llm/_torch/pyexecutor/scheduler/scheduler.py
• tests/integration/defs/kv_cache/test_prefix_aware_scheduling.py
• tests/integration/test_lists/test-db/l0_b200.yml
• tests/integration/test_lists/test-db/l0_h100.yml
• tests/unittest/_torch/executor/test_kvcache_aware_router.py
• tests/unittest/_torch/executor/test_py_scheduler.py

---

📝 Walkthrough

Walkthrough

The PR removes the Python-side caching layer for prefix-reuse summaries in the scheduler, simplifies related unit tests, deletes integration tests validating prefix-aware scheduling behavior, and updates test configurations to exclude these tests.

Changes

Cohort / File(s)	Summary
Core Implementation cpp/include/tensorrt_llm/batch_manager/templatedTrie.h, tensorrt_llm/_torch/pyexecutor/scheduler/scheduler.py	Removes prefix-reuse summary caching in Python scheduler and eliminates the node back-pointer reset in the C++ trie deletion path. Updates block accounting APIs to remove cached_summary parameters and adjusts draft-token presence checks from method calls to property access. Removes clamping logic for remaining_space calculation.
Integration Test Removal tests/integration/defs/kv_cache/test_prefix_aware_scheduling.py	Deletes the entire test suite that validated prefix-aware scheduling end-to-end behavior, including server launch, health checks, and metrics parsing.
Test Configuration Updates tests/integration/test_lists/test-db/l0_b200.yml, tests/integration/test_lists/test-db/l0_h100.yml	Removes prefix-aware scheduling test entries from GPU-tier test lists, reducing the number of executed integration tests for these configurations.
Unit Test Simplifications tests/unittest/_torch/executor/test_kvcache_aware_router.py, tests/unittest/_torch/executor/test_py_scheduler.py	Simplifies mock verification in routing tests by removing call-count assertions while retaining return value validation. Removes two unit tests covering draft-token edge cases in mixed batches and zero-draft scenarios.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch revert-13243-prefix-aware-tests-fixs

---

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