[None][fix] Allow fail-early when reuse block and legacy mamba cache

[Wanli-Jiang]

Summary by CodeRabbit

Release Notes

• Bug Fixes
  • Enhanced error handling for KV cache configuration: now provides clear guidance when incompatible block reuse settings are detected, directing users to the supported cache manager option.

Description

Test Coverage

PR Checklist

Please review the following before submitting your PR:

• PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.

• PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.

• Test cases are provided for new code paths (see test instructions)

• If PR introduces API changes, an appropriate PR label is added - either api-compatible or api-breaking. For api-breaking, include BREAKING in the PR title.

• Any new dependencies have been scanned for license and vulnerabilities

• CODEOWNERS updated if ownership changes

• Documentation updated as needed

• Update tava architecture diagram if there is a significant design change in PR.

• The reviewers assigned automatically/manually are appropriate for the PR.

• Please check this after reviewing the above items as appropriate for this PR.

GitHub Bot Help

To see a list of available CI bot commands, please comment /bot help.

[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: e0602af6-f0f6-40c0-ab09-5b59088d398b

📥 Commits

Reviewing files that changed from the base of the PR and between 31d4301 and 35a1a7d.

📒 Files selected for processing (1)

• tensorrt_llm/_torch/pyexecutor/_util.py

---

📝 Walkthrough

Walkthrough

In KVCacheCreator._get_model_kv_cache_manager_cls, the previous warn-only log for enable_block_reuse=True on hybrid linear models routed to the legacy path is replaced with a ValueError that fails fast when the resolved cache manager class is MixedMambaHybridCacheManager, directing users to CppMambaHybridCacheManager.

Changes

Block reuse guard for legacy hybrid cache manager

Layer / File(s)	Summary
ValueError guard replacing warn-only path tensorrt_llm/_torch/pyexecutor/_util.py	Removes the legacy warning log emitted when enable_block_reuse=True is combined with the disagg/env/speculative-decoding routing conditions, and replaces it with a ValueError raised whenever the resolved KV cache manager class is MixedMambaHybridCacheManager, with an error message instructing use of CppMambaHybridCacheManager.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

🚥 Pre-merge checks | ✅ 3 | ❌ 2

❌ Failed checks (2 warnings)

Check name	Status	Explanation	Resolution
Description check	⚠️ Warning	The PR description was not filled in; it contains only the template with no explanation of the issue, solution, test coverage, or implementation details.	Provide a description explaining the issue being fixed, the solution approach, and the test coverage for this change.
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (3 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly describes the main change: adding fail-early detection for the unsupported combination of block reuse and legacy mamba cache.
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.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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

[Wanli-Jiang]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #55222 [ run ] triggered by Bot. Commit: 20458ff Link to invocation

[nv-guomingz]

Let's await this #15533 ci pipeline

[tensorrt-cicd]

PR_Github #55222 [ run ] completed with state SUCCESS. Commit: 20458ff
/LLM/main/L0_MergeRequest_PR pipeline #44181 completed with status: 'FAILURE'

CI Report

⚠️ Action Required:

• Please check the failed tests and fix your PR
• If you cannot view the failures, ask the CI triggerer to share details
• Once fixed, request an NVIDIA team member to trigger CI again

Link to invocation

[Wanli-Jiang]

the change will be merged into #16597

[Wanli-Jiang]

actually we can drop this part since no longer used.