#Mac Mini M4 – Gateway stuck on “kerfuffling / pondering” with no error (local Ollama + Grok)

1 messages · Page 1 of 1 (latest)

steep tusk
#

Hey everyone,
Just got my Mac Mini M4 and I’m having the classic issue: gateway says connected but the agent never responds. It just sits on “kerfuffling...” or bouncing dots with no error, even after long waits.
Tried so far:

Multiple openclaw gateway restart
Full reset (rm -rf ~/.openclaw + openclaw onboard --install-daemon)
Reinstalling/restarting Ollama
Trying both dashboard and TUI
Deleting main agent folder
openclaw doctor --fix / --repair
Successfully added Grok (xAI) via API key, but still same issue whether using local Ollama or Grok

Has anyone on Mac Mini M4 found a reliable fix for this? Or is this a known compatibility issue with the current version on M4 hardware that might need a fix from the team?
Happy to provide logs or try suggested commands.
Thanks in advance!
Tim

zealous phoenixBOT
#

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

past jayBOT
half minnowBOT
#

I don’t think this smells like an M4-specific incompatibility by itself.

Because it hangs with both Ollama and Grok, I’d suspect the gateway/session run path more than the model backend.

Can you paste these, in this order:

openclaw --version
openclaw status --all
openclaw gateway status
openclaw models status
openclaw infer model run --prompt "Reply with exactly: smoke-ok" --json

Then, in another terminal, run:

# 
openclaw logs --follow

Send the log chunk from the moment you submit a message and it gets stuck.

What I’m specifically looking for:

  • Session file path must be within sessions directory
  • Cannot read properties of undefined (reading 'trim')
  • anything about bootstrapping / BOOTSTRAP.md
  • backend shape errors like messages[].content ... expected a string

Why this split helps:

  • if infer model run works, but chat/TUI still hangs, the model is fine and the session/gateway state is wedged
  • if infer model run also hangs, then it’s more likely provider/config/backend related
#

Two common culprits I’ve seen for this exact “connected but never replies” feel:

  1. stale sessions.json / broken session state
  2. leftover BOOTSTRAP.md keeping the agent in first-run mode

Docs:

I’d avoid more rm -rf ~/.openclaw for the moment, we’re past guessing and need the actual log signature. If you paste the outputs, I can help narrow it down fast.

#

That invite is not an official OpenClaw support path. Please ignore/report it.

Official support is here in #help / #help forum, and anything that needs maintainer attention should go to GitHub issues only:
https://github.com/openclaw/openclaw/issues

Tim, if you want, paste these and we can actually narrow the hang down:

openclaw --version
openclaw status --all
openclaw gateway status
openclaw models status
openclaw infer model run --prompt "Reply with exactly: smoke-ok" --json

Then grab the relevant chunk from: