Fix mp engine

[RunningLeon]

Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily receiving feedbacks. If you do not understand some items, don't worry, just make the pull request and seek help from maintainers.

Motivation

Please describe the motivation of this PR and the goal you want to achieve through this PR.

Modification

Please briefly describe what modification is made in this PR.

BC-breaking (Optional)

Does the modification introduce changes that break the backward-compatibility of the downstream repositories?
If so, please describe how it breaks the compatibility and how the downstream projects should modify their code to keep compatibility with this PR.

Use cases (Optional)

If this PR introduces a new feature, it is better to list some use cases here, and update the documentation.

Checklist

1. Pre-commit or other linting tools are used to fix the potential lint issues.
2. The modification is covered by complete unit tests. If not, please add more unit tests to ensure the correctness.
3. If the modification has a dependency on downstream projects of a newer version, this PR should be tested with all supported versions of downstream projects.
4. The documentation has been modified accordingly, like docstring or example tutorials.

[Copilot]

Pull request overview

This PR adjusts the multiprocessing (mp) engines (Ray + ZMQ) to better synchronize streaming session lifecycle, harden a few race/error cases, and improve observability around Ray startup.

Changes:

• Add session “ready/exist” signaling for streaming RPC paths (plumb an asyncio.Event through mp-engine streaming calls).
• Harden Ray shared actor creation and streaming output gathering against race/error cases.
• Add logging of Ray initialization address.

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
lmdeploy/pytorch/ray.py	Captures ray.init() context and logs the initialized address.
lmdeploy/pytorch/engine/mp_engine/zmq_rpc.py	Extends streaming client API to accept a session event and signal it once streaming begins.
lmdeploy/pytorch/engine/mp_engine/zmq_engine.py	Plumbs session event through ZMQ mp-engine streaming path.
lmdeploy/pytorch/engine/mp_engine/ray_engine.py	Adds init/error handling around Ray streaming tasks and plumbs session event through Ray mp-engine streaming path.
lmdeploy/pytorch/engine/mp_engine/base_worker.py	Avoids KeyError in EngineOutputGather.pop() when no gathered output exists for a stream.
lmdeploy/pytorch/engine/mp_engine/base.py	Changes streaming RPC abstract signature and uses a per-session asyncio.Event for lifecycle synchronization.
lmdeploy/pytorch/engine/engine_instance.py	Handles Ray actor creation race for shared store and introduces notify_add_msg_func hook in streaming infer.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.