#gateway issues

1 messages ยท Page 1 of 1 (latest)

modern ginkgoBOT
#

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

bronze canyonBOT
#

That combo usually means the service is installed/loaded, but the gateway process is crashing immediately (so the port stays free and the health-check never goes green).

Can you paste the output of these (in order)?

openclaw --version
openclaw gateway status
openclaw status --all
openclaw doctor

While youโ€™re at it, the money line is almost always in the logs right after a restart:

openclaw gateway restart
openclaw logs --follow
#

Things to look for in openclaw gateway status:

  • Last gateway error: (often points straight at the crash cause)
  • Config (cli) vs Config (service) mismatch (editing one config, service running another). If they differ, fix with:
    openclaw gateway install --force
    openclaw gateway restart
    

Docs runbook section that matches your symptom: https://docs.openclaw.ai/gateway/troubleshooting#gateway-service-not-running

viscid vine
#

OpenClaw 2026.3.8 (3caab92)

๐Ÿฆž OpenClaw 2026.3.8 (3caab92)
Shell yeahโ€”I'm here to pinch the toil and leave you the glory.

โ”‚
โ—‡ Doctor warnings โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ•ฎ
โ”‚ โ”‚
โ”‚ - channels.telegram.groupPolicy is "allowlist" but groupAllowFrom (and โ”‚
โ”‚ allowFrom) is empty โ€” all group messages will be silently dropped. โ”‚
โ”‚ Add sender IDs to channels.telegram.groupAllowFrom or โ”‚
โ”‚ channels.telegram.allowFrom, or set groupPolicy to "open". โ”‚
โ”‚ โ”‚
โ”œโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ•ฏ
โ”‚
โ—‡
Service: LaunchAgent (loaded)
File logs: /tmp/openclaw/openclaw-2026-03-11.log
Command: /opt/homebrew/opt/node/bin/node /opt/homebrew/Cellar/openclaw-cli/2026.3.2/libexec/lib/node_modules/openclaw/dist/index.js gateway --port 18789
Service file: ~/Library/LaunchAgents/ai.openclaw.gateway.plist
Service env: OPENCLAW_GATEWAY_PORT=18789

Service config looks out of date or non-standard.
Service config issue: Gateway service embeds OPENCLAW_GATEWAY_TOKEN and should be reinstalled. (Run openclaw gateway install --force to remove embedded service token.)
Recommendation: run "openclaw doctor" (or "openclaw doctor --repair").
Config (cli): ~/.openclaw/openclaw.json
Config (service): ~/.openclaw/openclaw.json

Gateway: bind=loopback (127.0.0.1), port=18789 (service args)
Probe target: ws://127.0.0.1:18789
Dashboard: http://127.0.0.1:18789/
Probe note: Loopback-only gateway; only local clients can connect.

Runtime: stopped (state active)
โ”‚

bronze canyonBOT
#

Yep, this explains it.

Your installed version is 2026.3.8, but the LaunchAgent is still trying to run the old Homebrew path:

/opt/homebrew/Cellar/openclaw-cli/2026.3.2/.../openclaw/dist/index.js ...

After a brew upgrade, that 2026.3.2 Cellar path is often gone โ†’ launchd starts it, it exits immediately โ†’ port stays free โ†’ health check times out.

Fix (do this first):

openclaw gateway install --force
openclaw gateway restart
openclaw gateway status

That also addresses the other warning:

#
  • โ€œGateway service embeds OPENCLAW_GATEWAY_TOKENโ€ฆโ€ โ†’ reinstall removes the embedded token from the plist so you donโ€™t get token drift later.

What โ€œgoodโ€ looks like after reinstall:

  • openclaw gateway status should show the Command pointing at 2026.3.8 (not 2026.3.2)
  • Runtime: running and RPC probe: ok

If it still stops immediately after that, paste the last ~100 lines of:

tail -n 120 /tmp/openclaw/openclaw-2026-03-11.log

(That Telegram groupPolicy allowlist warning is unrelated to the gateway not starting; it just means Telegram group messages would be dropped even when the gateway is healthy.)

viscid vine
#

okay its back up, now i need to connect my openclaw back to the alibaba cloud coding plan

bronze canyonBOT
#

If by โ€œAlibaba Cloud coding planโ€ you mean Alibaba Cloud Model Studio (DashScope) Coding Plan, the quickest way is to re-add that provider via the config wizard (so the LaunchAgent has everything it needs):

openclaw configure --section model

In the Model section, pick Alibaba Cloud Model Studio Coding Plan (Global/Intl or China), paste your API key, finish the wizard, then:

openclaw gateway restart
openclaw models status

Note (macOS/LaunchAgent gotcha): if you previously had the key only in your shell env, launchd wonโ€™t see it. Put it in ~/.openclaw/.env (env var is MODELSTUDIO_API_KEY) or enter it via openclaw configure so itโ€™s persisted correctly for the service.

If you meant Qwen OAuth free tier instead (also Alibaba/Qwen), say so and Iโ€™ll give the qwen-portal-auth steps.

viscid vine
#

okay i have it up now how do i give it read and exec permissions so it can read its old files?

bronze canyonBOT
#

Two separate things usually get mixed together here: (A) where the botโ€™s โ€œold filesโ€ live (workspace) and (B) whether the agent is allowed to use the read / exec tools.

1) Point the agent back at the old workspace (so the files are even visible)

Check what workspace itโ€™s using now:

openclaw config get agents.defaults.workspace

If your old files are in a different folder (common if you previously used something like ~/clawd or an older ~/.openclaw/workspace), set it back:

openclaw config set agents.defaults.workspace "/path/to/your/old/workspace"
openclaw gateway restart

2) Make sure the tools arenโ€™t restricted

#

If you (or onboarding) enabled a restrictive tool profile/allowlist, the agent may not be allowed to read/exec.

Check:

openclaw config get tools.profile
openclaw config get tools.allow
openclaw config get tools.deny

If you want a sane default that includes file read/write + exec tooling, set:

openclaw config set tools.profile "coding"
openclaw gateway restart
#

(That profile includes group:fs + group:runtime.)

3) If you specifically want exec to work on macOS

exec often needs to run on the gateway host (not a sandbox), and it may be gated by exec approvals/allowlists.

Fast test in chat (session-only):

/exec host=gateway security=allowlist ask=on-miss

Then manage approvals in Control UI โ†’ Nodes โ†’ Exec approvals (target โ€œGatewayโ€). Docs: https://docs.openclaw.ai/tools/exec-approvals


If you tell me where the old files are (path) and what error youโ€™re seeing (โ€œtool not allowedโ€ vs โ€œfile not foundโ€ vs โ€œpermission deniedโ€), I can point you to the exact knob to turn.

viscid vine
#

this didnt work: /exec host=gateway security=allowlist ask=on-miss