[Bug]: Windows LM Studio installation remains at 75% and is not detected after successful llmster install
维护者通常 5 天内回复
@ckelseynv 已经在做这个了。
开始于 2026年9月11日。
评估
这个 Issue 还没有评估数据。
描述
PAIR version or commit
PAIR desktop application 0.91.7
Affected component
Engine or model management
Environment
- OS: Microsoft Windows 11 Pro 10.0.26200, build 26200
- Architecture: x64
- GPU: NVIDIA RTX PRO 6000 Blackwell Workstation Edition
- GPU driver: 32.0.15.8216
- LM Studio CLI: 1.3.3, commit
07b7252 - llmster: 0.0.23-1
- Cluster size: 2
Steps to reproduce
- Have LM Studio installed on Windows with its server running on port 1234.
- Install and open PAIR 0.91.7.
- Select the LM Studio Download/install action.
- Observe progress reach 75% at the
installingstage. - Wait approximately 108 seconds.
- Stop the existing server with
lms server stop, then fully stop the daemon withlms daemon down. - Run the same official
https://lmstudio.ai/install.ps1script manually. - Confirm that it successfully installs llmster 0.0.23-1.
- Restart PAIR and retry the LM Studio installation.
This was reproduced three times, including after confirming that lms.exe was present on PATH and executable by the same Windows user.
Expected behavior
PAIR should detect or adopt the existing LM Studio installation, configure its managed backend on port 1235, and complete registration. If an existing daemon must be stopped first, PAIR should report that requirement instead of remaining at 75%.
Actual behavior
PAIR reaches 75% and later reports:
engine "lmstudio" was not detected after install
PAIR's proxy takes port 1234 and repeatedly checks its expected backend at 127.0.0.1:1235, but that backend is not registered or kept running. engine-state.json continues to contain only Ollama.
Running the official LM Studio installer manually succeeds:
Install completed at %USERPROFILE%\.cache\lm-studio\llmster\0.0.23-1.
Installation finished successfully! llmster is ready to launch.
Closing PAIR and restoring the original LM Studio service returns port 1234 and the existing models to normal operation.
Sanitized logs
engine:install-progress
engine=lmstudio percent=75 stage=installing
[nvpair-engine-manager] WARN engine error reported
id=engine-manager:install-failed:lmstudio
message="engine \"lmstudio\" was not detected after install"
engine:install-progress
engine=lmstudio
error="engine \"lmstudio\" was not detected after install"
percent=-1
stage=failed
[lmstudio-proxy] INFO local backend updated
engine=lmstudio host=127.0.0.1 port=1235 healthy=false
Confirmations
- I searched existing issues for duplicates.
- This is not a security vulnerability.
- I agree to follow the Code of Conduct.
- 主要语言
- Go
- 星标
- 1.6k
- 派生
- 266
- 平均合并
- 3 天 15 小时
- 30 天内合并 PR
- 23
环境准备
- 没有 Dockerfile 或 Docker Compose 文件
- 有 Pull Request 模板
- 阅读贡献指南
从这里开始
- 先读完整个 Issue，再读项目的贡献指南。
- 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
- Fork 仓库，在一个分支上完成修改。
- 提交 Pull Request，并在描述里引用这个 Issue 编号。
NVIDIA/Personal-AI-Router 的其他 Issue
-
enhancement
难度 4/5 一周以上 新手友好度 12/100
NVIDIA/Personal-AI-Router#162 ·
维护者通常 5 天内回复
-
enhancement
难度 4/5 3-5 天 新手友好度 25/100
NVIDIA/Personal-AI-Router#154 · 1 条评论 ·
维护者通常 5 天内回复
-
[Feature]: [Ollama] Alert the user about model pull failures可能已有人在做 @ckelseynv 于 2 天前认领。 未关闭enhancement
难度 3/5 1-2 天 新手友好度 56/100
NVIDIA/Personal-AI-Router#152 · 1 条评论 · 已指派 1 人 ·
维护者通常 5 天内回复
-
[Bug]: /v1/responses bypasses model-owner filtering in the Ollama proxy可能已有人在做 @ilaigold 于 3 天前认领。 未关闭
难度 3/5 半天 新手友好度 84/100
NVIDIA/Personal-AI-Router#146 ·
维护者通常 5 天内回复
-
Make model download cancellation asynchronous可能已有人在做 @ckelseynv 于 19 天前认领。 未关闭enhancement
NVIDIA/Personal-AI-Router#115 · 已指派 1 人 ·
维护者通常 5 天内回复
查看 NVIDIA/Personal-AI-Router 的全部 Issue
相似的 Issue
-
agent-butler-finding chore
难度 2/5 1-3 小时 新手友好度 78/100
jordansmall/spindrift#5128 ·
维护者通常 1 天内回复
-
难度 1/5 1 小时以内 新手友好度 84/100
维护者通常 2 天内回复
-
License GOtome under Apache 2.0 instead of AGPL-3.0可能已有人在做 @praetorianer777 今天认领。 未关闭area:workflow documentation
难度 2/5 1-3 小时 新手友好度 84/100
praetorianer777/GoTome#212 ·
维护者通常 1 天内回复
-
ci: pin Go version in Build and Unit Test workflows (>=1.26.6 floats to 1.27.2 and breaks golangci-lint)可能已有人在做 @leninmehedy 今天认领。 未关闭
难度 2/5 1-3 小时 新手友好度 87/100
hashgraph/solo-weaver#1255 ·
维护者通常 1 天内回复
-
area/infra backlog type/chore
难度 2/5 1-3 小时 新手友好度 76/100
praetorianer777/Stator#374 ·
维护者通常 1 天内回复