[GH-ISSUE #207] How to make a whole mooncake wheel package after make -j? #91

New issue

Closed

opened 2026-10-29 21:09:09 +00:00 by bsakaguchi · 4 comments

bsakaguchi commented

2026-10-29 21:09:09 +00:00

Owner

Originally created by @Huixxi on GitHub (Apr 3, 2025).
Original GitHub issue: https://github.com/kvcache-ai/Mooncake/issues/207

I tried python setup.py bdist_wheel within Mooncake/mooncake-wheel folder but it seems can not work.

Originally created by @Huixxi on GitHub (Apr 3, 2025). Original GitHub issue: https://github.com/kvcache-ai/Mooncake/issues/207 I tried `python setup.py bdist_wheel` within Mooncake/mooncake-wheel folder but it seems can not work.

bsakaguchi closed this issue

2026-10-29 21:09:09 +00:00

bsakaguchi commented

2026-10-29 21:09:10 +00:00

Author

Owner

@stmatengss commented on GitHub (Apr 3, 2025):

Hi, you can produce a wheel package according to the procedures in https://github.com/kvcache-ai/Mooncake/blob/main/.github/workflows/release.yaml. BTW, we provide a whl package in the release page (https://github.com/kvcache-ai/Mooncake/releases/download/v0.3.0-alpha/mooncake_transfer_engine-0.3.0a0-cp310-cp310-manylinux2014_x86_64.whl). But, you should make sure the server is running Ubuntu 22.04 and python 3.10)

@stmatengss commented on GitHub (Apr 3, 2025): Hi, you can produce a wheel package according to the procedures in https://github.com/kvcache-ai/Mooncake/blob/main/.github/workflows/release.yaml. BTW, we provide a whl package in the release page (https://github.com/kvcache-ai/Mooncake/releases/download/v0.3.0-alpha/mooncake_transfer_engine-0.3.0a0-cp310-cp310-manylinux2014_x86_64.whl). But, you should make sure the server is running Ubuntu 22.04 and python 3.10)

bsakaguchi commented

2026-10-29 21:09:10 +00:00

Author

Owner

@Huixxi commented on GitHub (Apr 14, 2025):

Does mooncake+vllm still need etcd metadata server and a master server now?

@Huixxi commented on GitHub (Apr 14, 2025): Does mooncake+vllm still need etcd metadata server and a master server now?

bsakaguchi commented

2026-10-29 21:09:10 +00:00

Author

Owner

@stmatengss commented on GitHub (Apr 16, 2025):

Does mooncake+vllm still need etcd metadata server and a master server now?

You now have flexible connection options! Choose between using a metadata server or establishing direct handshake connections. Experience these new features in our latest update.

@stmatengss commented on GitHub (Apr 16, 2025): > Does mooncake+vllm still need etcd metadata server and a master server now? You now have flexible connection options! Choose between using a metadata server or establishing direct handshake connections. Experience these new features in our latest update.

bsakaguchi commented

2026-10-29 21:09:10 +00:00

Author

Owner

@Huixxi commented on GitHub (Apr 17, 2025):

Hi~ I see mooncake+vllm supports dynamic adding instance into a running server(even though I met some hang problem and seems doesn't work that stable https://github.com/kvcache-ai/Mooncake/blob/main/doc/en/vllm-integration-v1.md#run-example). But what I want more to ask is whether it supports dynamic remove instances from a running server?

@Huixxi commented on GitHub (Apr 17, 2025): Hi~ I see mooncake+vllm supports dynamic adding instance into a running server(even though I met some hang problem and seems doesn't work that stable https://github.com/kvcache-ai/Mooncake/blob/main/doc/en/vllm-integration-v1.md#run-example). But what I want more to ask is whether it supports dynamic remove instances from a running server?

bsakaguchi referenced this issue

2026-10-29 21:26:20 +00:00

[PR #91] [CLOSED] xx #1319