Add default set env MC_STORE_MEMCPY is 0 when protocol is tcp

[lengrongfu]

Current Mooncake version has a bug with TCP-only hosts causing a SEGFAULT error. Set MC_STORE_MEMCPY=0 until the kvcache-ai/Mooncake#1986 is fixed.

Although this documentation provides a hint, we can enhance usability by setting this default environment variable directly within the code.

If a user overlooks the guidance provided in this documentation, it would be extremely difficult to pinpoint the root cause of the issue based solely on the error message.

(raylet) A worker died or was killed while executing a task by an unexpected system error. To troubleshoot the problem, check the logs for the dead worker. Lease ID: 0b00000043c351dfa676154eb6e2d2d83b12cc84d4b0dd6f7c95895706e201b2 Worker ID: e80b5521d830d80614c25bcf5227a307abb4372b9112056245654785 Node ID: 881e316ecdb052ae3a7b7405e2cab0cc775044e5492e1473e235f653 Worker IP address: 10.233.74.188 Worker port: 35501 Worker PID: 887276 Worker exit type: SYSTEM_ERROR Worker exit detail: Worker unexpectedly exits with a connection error code 2. End of file. Some common causes include: (1) the process was killed by the OOM killer due to high memory usage, (2) ray stop --force was called, or (3) the worker crashed unexpectedly due to SIGSEGV or another unexpected error.
[2026-05-18 03:12:01,209] loop.py:130 INFO Shutting down 1 inference engine(s)...
Traceback (most recent call last):
  File "<frozen runpy>", line 198, in _run_module_as_main
  File "<frozen runpy>", line 88, in _run_code
  File "/home/jovyan/code/TorchSpec/torchspec/train_entry.py", line 416, in <module>
    train_async_no_generation(args)
  File "/home/jovyan/code/TorchSpec/torchspec/train_entry.py", line 403, in train_async_no_generation
    run_training_loop(
  File "/home/jovyan/code/TorchSpec/torchspec/controller/loop.py", line 470, in run_training_loop
    return training_loop(
           ^^^^^^^^^^^^^^
  File "/home/jovyan/code/TorchSpec/torchspec/controller/loop.py", line 314, in training_loop
    train_results = ray.get(train_futures)
                    ^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jovyan/code/TorchSpec/.venv/lib/python3.11/site-packages/ray/_private/auto_init_hook.py", line 22, in auto_init_wrapper
    return fn(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^
  File "/home/jovyan/code/TorchSpec/.venv/lib/python3.11/site-packages/ray/_private/client_mode_hook.py", line 107, in wrapper
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/home/jovyan/code/TorchSpec/.venv/lib/python3.11/site-packages/ray/_private/worker.py", line 2980, in get
    values, debugger_breakpoint = worker.get_objects(
                                  ^^^^^^^^^^^^^^^^^^^
  File "/home/jovyan/code/TorchSpec/.venv/lib/python3.11/site-packages/ray/_private/worker.py", line 1025, in get_objects
    raise value
ray.exceptions.ActorDiedError: The actor died unexpectedly before finishing this task.
        class_name: TrainerActor
        actor_id: 3b496743efbcb5868b136aba01000000
        pid: 887276
        namespace: 4b085ef6-2e8d-467e-b0b2-edf27331584d
        ip: 10.233.74.188
The actor is dead because its worker process has died. Worker exit type: SYSTEM_ERROR Worker exit detail: Worker unexpectedly exits with a connection error code 2. End of file. Some common causes include: (1) the process was killed by the OOM killer due to high memory usage, (2) ray stop --force was called, or (3) the worker crashed unexpectedly due to SIGSEGV or another unexpected error.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c81cdf697d

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".