[https://nvbugs/6426847][fix] Invert the check to `not isinstance(self.mpi_session, (MpiCommSession…

[trtllm-agent]

Summary

• Root cause: session_reuse test hook monkey-patches MpiPoolSession in tensorrt_llm.executor.proxy with a factory function, so isinstance(self.mpi_session, MpiPoolSession) raises TypeError: isinstance() arg 2 must be a type because the second arg is a function.
• Fix: Invert the check to not isinstance(self.mpi_session, (MpiCommSession, RemoteMpiCommSessionClient)) — those two types are never monkey-patched and their absence uniquely identifies the local pool-session case (same fix pattern used earlier in the same function).
• 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/6426847

Summary by CodeRabbit

• Bug Fixes
  • Improved worker process monitoring across supported execution session types.
  • Ensured worker identities are registered consistently when initialization provides the expected status information.

[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: 068cf9b1-282d-4bbd-bb69-670d14929b49

📥 Commits

Reviewing files that changed from the base of the PR and between 3b330ca and 87aec0e.

📒 Files selected for processing (1)

• tensorrt_llm/executor/proxy.py

---

📝 Walkthrough

Walkthrough

The executor worker startup logic now extracts and registers worker process identities for any three-element initialization status unless the MPI session is a communication session or remote communication client.

Changes

Worker identity registration

Layer / File(s)	Summary
Broaden worker identity registration guard tensorrt_llm/executor/proxy.py	Worker process identities are registered when initialization returns a three-element status for MPI sessions other than MpiCommSession and RemoteMpiCommSessionClient.

Estimated code review effort: 2 (Simple) | ~5 minutes

Possibly related PRs

• NVIDIA/TensorRT-LLM#16338: Updates related worker identity extraction and registration in GenerationExecutorProxy.
• NVIDIA/TensorRT-LLM#16411: Changes the same worker identity registration condition in GenerationExecutorProxy.

Suggested reviewers: cascade812, tongyuantongyu, shixiaowei02

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title matches the PR’s main fix and follows the required [ticket][type] pattern.
Description check	✅ Passed	The description covers the bug, fix, tests, and link, so it is mostly complete despite template section name mismatches.
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.}

[QiJune]

/bot run

[tensorrt-cicd]

PR_Github #59436 [ run ] triggered by Bot. Commit: 87aec0e Link to invocation

[tensorrt-cicd]

PR_Github #59436 [ run ] completed with state SUCCESS. Commit: 87aec0e
/LLM/main/L0_MergeRequest_PR pipeline #47905 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

CI Agent Failure Analysis

Link to invocation

[chienchunhung]

FYI I am preparing PR#16444 as the fix and test unwaiving.

[BowenFu]

Heads up: this is one of three competing fixes for the same one-line session check in proxy.py — #16417, #16490, #16496 (they touch the same line and would conflict; only one should land). #16490 uses the polymorphic self.mpi_session.is_comm_session(), which resolves the class refs inside mpi_session's namespace and is immune to the test-infra rebinding of the module-level MpiPoolSession symbol (the root cause). Suggest consolidating on #16490 and closing this one, folding its nvbug un-waive into #16490.

[brnguyen2]

Superseded by #16444, which has merged.