[https://nvbugs/6463828][fix] Replace the concrete-type check with the polymorphic…

[trtllm-agent]

Summary

• Root cause: isinstance(self.mpi_session, MpiPoolSession) raised TypeError because the test-suite monkey-patch in tests/test_common/session_reuse.py replaces tensorrt_llm.executor.proxy.MpiPoolSession with a factory function (not a type), and the actual session is a _ReusableSession wrapper (not a subclass) that delegates via __getattr__.
• Fix: Replace the concrete-type check with the polymorphic MpiSession.is_comm_session() — returns False for MpiPoolSession (register PIDs), True for comm-based sessions (skip). The _ReusableSession wrapper transparently forwards the method to its real MpiPoolSession via __getattr__.
• Automated fix generated by repair-bot

Test plan

• Verify fix on the same GPU type as the original failure
• Check for regressions in related tests

Links

• Bug: https://nvbugs/6463828

Summary by CodeRabbit

• Bug Fixes
  • Improved worker process tracking during startup across different MPI session types.
  • Ensured operating system process identities are registered consistently when available.

[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: 60a6ad38-cc2b-4ce9-a2e4-27ee48574640

📥 Commits

Reviewing files that changed from the base of the PR and between 4277d15 and a108615.

📒 Files selected for processing (1)

• tensorrt_llm/executor/proxy.py

---

📝 Walkthrough

Walkthrough

GenerationExecutorProxy now registers worker process identities for any non-communication MPI session when initialization returns identity data, replacing the previous MpiPoolSession-specific check.

Changes

MPI worker identity registration

Layer / File(s)	Summary
Session-aware worker registration tensorrt_llm/executor/proxy.py	Worker identity registration now checks not self.mpi_session.is_comm_session() alongside len(status) == 3, with comments documenting communication-session monitoring limitations.

Estimated code review effort: 1 (Trivial) | ~5 minutes

Possibly related PRs

• NVIDIA/TensorRT-LLM#16491: Applies the same session-aware condition in the worker initialization path.
• NVIDIA/TensorRT-LLM#16490: Changes the same worker identity registration predicate.
• NVIDIA/TensorRT-LLM#16444: Updates the same registration logic and extracts worker registration into a helper.

Suggested reviewers: asfiyab-nvidia

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title matches the main change: replacing a concrete MPI session type check with a polymorphic helper.
Description check	✅ Passed	The description covers the root cause, fix, and test plan, but it omits the template's explicit PR checklist section.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
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.}