leaden drift
Hi @stray oxide
I'm installing OpenClaw on windows using npm install command. I'm using a local gateway and loopback bind. As for the model, I'm using LM Studio Gemma 4 E2B with 32k context length.
When I send a message, OpenClaw is shown loading a response but no activity is occuring in LM Studio. Below are partial logs from openclaw gateway --verbose:
12:14:04 [diagnostic] stuck session: sessionId=unknown sessionKey=agent:main:main state=processing age=133s queueDepth=1
12:21:08 [ws] closed before connect conn=79cc0aed-626e-48cd-ad33-c783cd48aff4 peer=127.0.0.1:51040->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=n/a host=127.0.0.1:18789 ua=n/a code=1006 reason=n/a
12:21:10 [agent/embedded] embedded run prompt start: runId=6af9800d-ce06-46eb-b2ec-9183989e0a36 sessionId=ff7ab951-ef37-4ec6-8b5f-26cb7f04ac2f provider=lmstudio api=openai-completions endpoint=local route=local policy=none
12:21:10 [agent/embedded] [context-diag] pre-prompt: sessionKey=agent:main:main messages=0 roleCounts=none historyTextChars=0 maxMessageTextChars=0 historyImageBlocks=0 systemPromptChars=28418 promptChars=287 promptImages=0 provider=lmstudio/google/gemma-4-e2b sessionFile=C:\Users\Lim Wei Xiang.openclaw\agents\main\sessions\ff7ab951-ef37-4ec6-8b5f-26cb7f04ac2f.jsonl
12:21:10 [agent/embedded] embedded run agent start: runId=6af9800d-ce06-46eb-b2ec-9183989e0a36
12:21:11 [agent/embedded] embedded run agent end: runId=6af9800d-ce06-46eb-b2ec-9183989e0a36 isError=true model=google/gemma-4-e2b provider=lmstudio error=LLM request failed: provider rejected the request schema or tool payload. rawError=400 Invalid 'reasoning_effort' value: 'on'. Supported values: none, minimal, low, medium, high, xhigh.