openclaw stuck on booting gateway | Friends of the Crustacean 🦞🤝 | Page 1

broken rover Apr 28, 2026, 3:39 PM

#

after updating to latest version 4.26 my openclaw is stuck on 'starting channels and sidecars'

stiff minnowBOT Apr 28, 2026, 3:40 PM

#

That usually means Gateway reached HTTP startup, then wedged while starting channel providers / plugin sidecars.

First, can you paste these?

openclaw --version
node -v
openclaw status --all
openclaw gateway status

Then try the safe first pass:

openclaw doctor
openclaw gateway restart

#

openclaw logs --follow

If it hangs again at starting channels and sidecars, isolate channels:

openclaw config set channels.telegram.enabled false --strict-json
openclaw config set channels.whatsapp.enabled false --strict-json
openclaw gateway restart

Only run those disables if you’re not relying on Telegram/WhatsApp right now — it’s just to identify whether one channel is blocking startup.

Docs for the debug flow: https://docs.openclaw.ai/help/faq and https://docs.openclaw.ai/gateway/troubleshooting

broken rover Apr 28, 2026, 3:47 PM

#

message is going to take multiple posts to submit. i'll submit them one at a time. i'll let you know whe nthey're all submitted

stiff minnowBOT Apr 28, 2026, 3:47 PM

#

Sounds good — send them in chunks. I’ll wait until you say they’re all submitted before digging in, so we don’t chase a half-log.

broken rover Apr 28, 2026, 3:49 PM

#

C:\Users\Eagle3dStreaming> openclaw --version
OpenClaw 2026.4.26 (be8c246)

C:\Users\Eagle3dStreaming> node -v
v22.22.0

C:\Users\Eagle3dStreaming> openclaw status --all

🦞 OpenClaw 2026.4.26 (be8c246) — Type the command with confidence—nature will provide the stack trace if needed.

Overview

Version is 2026.4.26. OS is windows 10.0.19045 (x64). Node is 22.22.0. Config is at ~.openclaw\openclaw.json. Dashboard is at http://127.0.0.1:18789/. Tailscale exposure is off. Channel is stable (default). Update status: pnpm, up to date, npm latest 2026.4.26. Gateway is local at ws://127.0.0.1:18789 (local loopback) and is unreachable (connect ECONNREFUSED 127.0.0.1:18789). Security: run openclaw security audit --deep. Gateway self is unknown. Gateway service is a Scheduled Task, installed and registered, but stopped (state Ready, Task Last Run Result=1; treating as not running). Node service is a Scheduled Task, not installed. Agents: 4 total, 2 bootstrapping, 0 active, 13 sessions. Secrets: none.

Channels

No channels listed.

Agents

main — bootstrap file PRESENT, 1 session, last active 5d ago, store at ~.openclaw\agents\main\sessions\sessions.json.

filings-agent (Filings Agent) — bootstrap file PRESENT, 6 sessions, last active 1h ago, store at ~.openclaw\agents\filings-agent\sessions\sessions.json.

financial-statements-agent (Financial Statements Agent) — bootstrap file ABSENT, 5 sessions, last active 27m ago, store at ~.openclaw\agents\financial-statements-agent\sessions\sessions.json.

intrinsic-value-agent (Intrinsic Value Agent) — bootstrap file ABSENT, 1 session, last active 12h ago, store at ~.openclaw\agents\intrinsic-value-agent\sessions\sessions.json.

Diagnosis (read-only)

Gateway connection details: target ws://127.0.0.1:18789, source local loopback, config C:\Users\Eagle3dStreaming.openclaw\openclaw.json, bind loopback.

stiff minnowBOT Apr 28, 2026, 3:49 PM

#

Got it — first chunk received. Keep going, and I’ll wait for your “all submitted” before analyzing.

One quick note I’m already seeing: the gateway service is currently stopped with Scheduled Task result 1, so the key thing will be the gateway status/log output showing why it exited.

broken rover Apr 28, 2026, 3:49 PM

#

✓ Config: C:\Users\Eagle3dStreaming.openclaw\openclaw.json
✓ Secret diagnostics (0)
! Restart sentinel present — gateway restart update skipped (unknown) 28m ago
✓ Port 18789
! Tailscale exposure: off, daemon unknown
✓ Skills: 7 eligible, 0 missing, C:\Users\Eagle3dStreaming.openclaw\workspace
✓ Plugin compatibility (none)
Gateway restart attempts (tail) at C:\Users\Eagle3dStreaming.openclaw\logs\gateway-restart.log:
[Thu 04/23/2026 11:17:55.80] openclaw restart log initialized
[Thu 04/23/2026 11:17:55.82] openclaw restart attempt source=windows-task-handoff target="OpenClaw Gateway"
[Thu 04/23/2026 11:17:57.23] openclaw restart finished source=windows-task-handoff
[Tue 04/28/2026 8:24:55.22] openclaw restart log initialized
[Tue 04/28/2026 8:24:55.22] openclaw restart attempt source=update target=OpenClaw Gateway
! Channel issues skipped (gateway unreachable)
Gateway health: connect ECONNREFUSED 127.0.0.1:18789
Pasteable debug report. Auth tokens redacted. Troubleshooting: https://docs.openclaw.ai/troubleshooting
C:\Users\Eagle3dStreaming> openclaw gateway status
🦞 OpenClaw 2026.4.26 (be8c246) — iMessage green bubble energy, but for everyone.
Service: Scheduled Task (registered). File logs at C:\Users\EAGLE3~1\AppData\Local\Temp\openclaw\openclaw-2026-04-28.log. Command: C:\Program Files\nodejs\node.exe C:\Users\Eagle3dStreaming\AppData\Roaming\npm\node_modules\openclaw\dist\index.js gateway --port 18789. Service file: ~.openclaw\gateway.cmd. Service env: OPENCLAW_GATEWAY_PORT=18789.
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.

stiff minnowBOT Apr 28, 2026, 3:49 PM

#

Received chunk 2. I’ll keep holding until you say it’s complete.

The next most useful bit will be the tail of:

type C:\Users\EAGLE3~1\AppData\Local\Temp\openclaw\openclaw-2026-04-28.log

or, if it’s huge, just the last ~100 lines.

broken rover Apr 28, 2026, 3:49 PM

#

Runtime: stopped (state Ready, last run 1, last run time 4/28/2026 8:36:10 AM, Task Last Run Result=1; treating as not running). Connectivity probe: failed. Probe target ws://127.0.0.1:18789, connect ECONNREFUSED 127.0.0.1:18789. Capability: unknown.
Service is loaded but not running (likely exited immediately). File logs: C:\Users\EAGLE3~1\AppData\Local\Temp\openclaw\openclaw-2026-04-28.log. Logs: schtasks /Query /TN "OpenClaw Gateway" /V /FO LIST. Restart attempts: C:\Users\Eagle3dStreaming.openclaw\logs\gateway-restart.log.
Troubles: run openclaw status. Troubleshooting: https://docs.openclaw.ai/troubleshooting.

now it is complete

stiff minnowBOT Apr 28, 2026, 3:50 PM

#

Thanks — the important bit is: this is not currently stuck; the Scheduled Task is exiting immediately:

Runtime: stopped ... Task Last Run Result=1

So we need the actual file log, not more status output.

Please paste the last ~150 lines of this:

Get-Content "$env:TEMP\openclaw\openclaw-2026-04-28.log" -Tail 150

If that log doesn’t show the crash clearly, run the gateway in the foreground so Windows prints the real error:

set OPENCLAW_GATEWAY_STARTUP_TRACE=1

#

openclaw gateway --port 18789

Leave that running until it either exits or hangs, then paste the last output. That should tell us whether this is channel startup, plugin deps, config, or a Windows Scheduled Task/service issue.

broken rover Apr 28, 2026, 3:51 PM

#

C:\Users\Eagle3dStreaming>Get-Content "$env:TEMP\openclaw\openclaw-2026-04-28.log" -Tail 150
'Get-Content' is not recognized as an internal or external command,
operable program or batch file.

C:\Users\Eagle3dStreaming>

stiff minnowBOT Apr 28, 2026, 3:51 PM

#

You’re in cmd, not PowerShell — my bad. Use this cmd version:

powershell -NoProfile -Command "Get-Content \"$env:TEMP\openclaw\openclaw-2026-04-28.log\" -Tail 150"

Or plain cmd:

type C:\Users\EAGLE3~1\AppData\Local\Temp\openclaw\openclaw-2026-04-28.log

If it’s too long, the PowerShell one is better because it tails only the end.

broken rover Apr 28, 2026, 3:52 PM

#

{"0":"{"subsystem":"gateway"}","1":"log file: C:\Users\EAGLE3~1\AppData\Local\Temp\openclaw\openclaw-2026-04-28.log","_meta":{"runtime":"node","runtimeVersion":"22.22.0","hostname":"MP15-MS05","name":"{"subsystem":"gateway"}","parentNames":["openclaw"],"date":"2026-04-28T15:37:04.769Z","logLevelId":3,"logLevelName":"INFO","path":{"fullFilePath":"file:///C:/Users/Eagle3dStreaming/AppData/Roaming/npm/node_modules/openclaw/dist/subsystem-rHhUC6qs.js:147:14","fileName":"subsystem-rHhUC6qs.js","fileNameWithLine":"subsystem-rHhUC6qs.js:147","fileColumn":"14","fileLine":"147","filePath":"/C:/Users/Eagle3dStreaming/AppData/Roaming/npm/node_modules/openclaw/dist/subsystem-rHhUC6qs.js","filePathWithLine":"/C:/Users/Eagle3dStreaming/AppData/Roaming/npm/node_modules/openclaw/dist/subsystem-rHhUC6qs.js:147","method":"logToFile"}},"time":"2026-04-28T08:37:04.769-07:00","hostname":"MP15-MS05","message":"log file: C:\Users\EAGLE3~1\AppData\Local\Temp\openclaw\openclaw-2026-04-28.log"}
{"0":"{"subsystem":"gateway"}","1":"starting channels and sidecars...","_meta":{"runtime":"node","runtimeVersion":"22.22.0","hostname":"MP15-MS05","name":"{"subsystem":"gateway"}","parentNames":["openclaw"],"date":"2026-04-28T15:37:04.849Z","logLevelId":3,"logLevelName":"INFO","path":{"fullFilePath":"file:///C:/Users/Eagle3dStreaming/AppData/Roaming/npm/node_modules/openclaw/dist/subsystem-rHhUC6qs.js:147:14","fileName":"subsystem-rHhUC6qs.js","fileNameWithLine":"subsystem-rHhUC6qs.js:147","fileColumn":"14","fileLine":"147","filePath":"/C:/Users/Eagle3dStreaming/AppData/Roaming/npm/node_modules/openclaw/dist/subsystem-rHhUC6qs.js","filePathWithLine":"/C:/Users/Eagle3dStreaming/AppData/Roaming/npm/node_modules/openclaw/dist/subsystem-rHhUC6qs.js:147","method":"logToFile"}},"time":"2026-04-28T08:37:04.849-07:00","hostname":"MP15-MS05","message":"starting channels and sidecars..."}

stiff minnowBOT Apr 28, 2026, 3:53 PM

#

Yep, that log is too sparse: it only confirms Gateway reaches sidecars, then never reaches gateway ready.

Run this in cmd to bypass the Scheduled Task and get startup timings:

set OPENCLAW_GATEWAY_STARTUP_TRACE=1
openclaw gateway --port 18789

Wait ~60s. If it exits or hangs, paste the output.

Then run the isolation test:

set OPENCLAW_GATEWAY_STARTUP_TRACE=1
set OPENCLAW_SKIP_CHANNELS=1

#

openclaw gateway --port 18789

Interpretation:

If the second one reaches gateway ready → one of the channels/model prewarm is blocking startup.
If it still hangs/exits → it’s likely plugin sidecars/ACP/runtime deps, not channels.

After that we’ll know which path to fix. Docs: https://docs.openclaw.ai/gateway/troubleshooting

broken rover Apr 28, 2026, 3:56 PM

#

it hung, with this output

C:\Users\Eagle3dStreaming>openclaw gateway --port 18789

🦞 OpenClaw 2026.4.26 (be8c246) — More integrations than your therapist's intake form.

08:53:54 [gateway] startup trace: runtime.config 4.5ms total=1978.5ms eventLoopMax=0.0ms
08:53:54 [gateway] startup trace: control-ui.root 5.3ms total=1989.4ms eventLoopMax=0.0ms
08:53:54 [gateway] startup trace: tls.runtime 0.3ms total=1993.9ms eventLoopMax=0.0ms
08:53:54 [gateway] starting HTTP server...
08:53:54 [canvas] host mounted at http://127.0.0.1:18789/__openclaw__/canvas/ (root C:\Users\Eagle3dStreaming.openclaw\canvas)
08:53:54 [gateway] startup trace: runtime.state 13.2ms total=2015.2ms eventLoopMax=0.0ms
08:53:54 [gateway] startup trace: runtime.early 148.1ms total=2171.3ms eventLoopMax=0.0ms
08:53:54 [health-monitor] started (interval: 300s, startup-grace: 60s, channel-connect-grace: 120s)
08:53:55 [gateway] startup trace: http.bound 205.4ms total=2376.6ms eventLoopMax=0.0ms
08:53:55 [gateway] startup trace: post-attach.update-sentinel 133.9ms total=2515.0ms eventLoopMax=9.7ms
08:53:55 [gateway] agent model: anthropic/claude-sonnet-4-6
08:53:55 [gateway] http server listening (0 plugins, 5.1s)
08:53:55 [gateway] log file: C:\Users\EAGLE3~1\AppData\Local\Temp\openclaw\openclaw-2026-04-28.log
08:53:55 [gateway] startup trace: post-attach.log 81.5ms total=2600.6ms eventLoopMax=0.0ms
08:53:55 [gateway] startup trace: post-attach.update-check 3.8ms total=2610.3ms eventLoopMax=0.0ms
08:53:55 [gateway] starting channels and sidecars...
08:53:55 [gateway] startup trace: sidecars.session-locks 9.6ms total=2629.0ms eventLoopMax=0.0ms
08:53:55 [gateway] startup trace: sidecars.gmail-watch 0.1ms total=2633.1ms eventLoopMax=0.0ms
08:53:55 [gateway] startup trace: sidecars.gmail-model 0.2ms total=2636.7ms eventLoopMax=0.0ms
08:53:55 [gateway] startup trace: sidecars.internal-hooks 0.2ms total=2640.8ms eventLoopMax=0.0ms

stiff minnowBOT Apr 28, 2026, 3:57 PM

#

Good, that pinpoints it: it hangs right after sidecars.internal-hooks, which means the next phase is channel startup / model prewarm.

Please stop that foreground process with Ctrl+C, then run exactly this in the same cmd window:

set OPENCLAW_GATEWAY_STARTUP_TRACE=1
set OPENCLAW_SKIP_CHANNELS=1
openclaw gateway --port 18789

If that reaches gateway ready, then we know the core Gateway is fine and one channel/provider startup is wedging.

If it works, next quick fix is to disable unused channels and restart normally, e.g.:

openclaw config set channels.telegram.enabled false --strict-json

#

openclaw config set channels.whatsapp.enabled false --strict-json
openclaw gateway restart

Paste the OPENCLAW_SKIP_CHANNELS=1 result first though — that’s the fork in the road.

broken rover Apr 28, 2026, 3:59 PM

#

08:58:45 [gateway] agent model: anthropic/claude-sonnet-4-6
08:58:45 [gateway] http server listening (0 plugins, 4.9s)
08:58:45 [gateway] log file: C:\Users\EAGLE3~1\AppData\Local\Temp\openclaw\openclaw-2026-04-28.log
08:58:45 [gateway] startup trace: post-attach.log 80.9ms total=2539.9ms eventLoopMax=0.0ms
08:58:45 [gateway] startup trace: post-attach.update-check 3.1ms total=2546.9ms eventLoopMax=0.0ms
08:58:45 [gateway] starting channels and sidecars...
08:58:45 [gateway] startup trace: sidecars.session-locks 14.3ms total=2569.6ms eventLoopMax=0.0ms
08:58:45 [gateway] startup trace: sidecars.gmail-watch 0.1ms total=2573.8ms eventLoopMax=0.0ms
08:58:45 [gateway] startup trace: sidecars.gmail-model 0.2ms total=2577.6ms eventLoopMax=0.0ms
08:58:45 [gateway] startup trace: sidecars.internal-hooks 0.2ms total=2581.4ms eventLoopMax=0.0ms
08:58:45 [gateway/channels] skipping channel start (OPENCLAW_SKIP_CHANNELS=1 or OPENCLAW_SKIP_PROVIDERS=1)
08:58:45 [gateway] startup trace: sidecars.channels 4.1ms total=2589.6ms eventLoopMax=0.0ms
08:58:45 [gateway] startup trace: sidecars.plugin-services 2.8ms total=2598.1ms eventLoopMax=0.0ms
08:58:45 [gateway] startup trace: sidecars.memory 0.2ms total=2602.4ms eventLoopMax=0.0ms
08:58:45 [gateway] startup trace: sidecars.restart-sentinel 18.1ms total=2624.7ms eventLoopMax=0.0ms
08:58:45 [gateway] startup trace: sidecars.subagent-recovery 5.3ms total=2634.1ms eventLoopMax=0.0ms
08:58:45 [gateway] startup trace: sidecars.main-session-recovery 59.3ms total=2705.3ms eventLoopMax=0.0ms
08:58:45 [gateway] startup trace: sidecars.total 156.6ms total=2711.2ms eventLoopMax=0.0ms
08:58:45 [gateway] startup trace: sidecars.ready 5.0ms total=2716.3ms eventLoopMax=0.0ms
08:58:45 [gateway] ready
08:58:45 [gateway] startup trace: runtime.post-attach 414.4ms total=2725.5ms eventLoopMax=0.0ms
08:58:45 [gateway] startup trace: ready 3.7ms total=2729.2ms eventLoopMax=0.0ms
08:58:45 [heartbeat] started

stiff minnowBOT Apr 28, 2026, 3:59 PM

#

Perfect — core Gateway is healthy. The hang is definitely in channel startup.

Now isolate the offending channel:

Stop the foreground gateway: Ctrl+C
Print configured channel names without tokens:

node -e "const c=require(process.env.USERPROFILE+'/.openclaw/openclaw.json'); for (const [k,v] of Object.entries(c.channels||{})) console.log(k, 'enabled=', v&&v.enabled)"

Temporarily disable channels one at a time, then retry normal startup:

openclaw config set channels.telegram.enabled false --strict-json

#

openclaw config set channels.whatsapp.enabled false --strict-json
openclaw gateway restart

If you have other enabled channels from step 2, disable those too with the same pattern:

openclaw config set channels.CHANNELNAME.enabled false --strict-json

Once Gateway starts normally, re-enable channels one by one to find the culprit:

openclaw config set channels.telegram.enabled true --strict-json
openclaw gateway restart

#

My bet is one channel provider is blocking during connect/polling after the update, not the Gateway itself.

broken rover Apr 28, 2026, 4:06 PM

#

🦞 OpenClaw 2026.4.26 (be8c246) — Your terminal just grew claws—type something and let the bot pinch the busywork.

chokidar@^5.0.0
09:03:22 [gateway] starting HTTP server...
09:03:22 [canvas] host mounted at http://127.0.0.1:18789/__openclaw__/canvas/ (root C:\Users\Eagle3dStreaming.openclaw\canvas)
09:03:24 [health-monitor] started (interval: 300s, startup-grace: 60s, channel-connect-grace: 120s)
09:03:25 [gateway] agent model: anthropic/claude-sonnet-4-6
09:03:25 [gateway] http server listening (7 plugins: acpx, bonjour, browser, device-pair, memory-core, phone-control, talk-voice; 39.5s)
09:03:25 [gateway] log file: C:\Users\EAGLE3~1\AppData\Local\Temp\openclaw\openclaw-2026-04-28.log
09:03:26 [gateway] starting channels and sidecars...
09:03:29 [plugins] bonjour: watchdog detected non-announced service; attempting re-advertise (gateway fqdn=MP15-MS05 (OpenClaw) (2)._openclaw-gw._tcp.local. host=MP15-MS05-(2).local. port=18789 state=probing)
09:04:54 [plugins] bonjour: restarting advertiser (service stuck in probing for 85045ms (gateway fqdn=MP15-MS05 (OpenClaw) (2)._openclaw-gw._tcp.local. host=MP15-MS05-(2).local. port=18789 state=probing))
09:04:54 [plugins] bonjour: advertised gateway fqdn=MP15-MS05 (OpenClaw) (2)._openclaw-gw._tcp.local. host=MP15-MS05-(2).local. port=18789 state=unannounced
09:04:54 [plugins] bonjour: advertised gateway fqdn=MP15-MS05 (OpenClaw)._openclaw-gw._tcp.local. host=MP15-MS05.local. port=18789 state=unannounced
09:04:56 [model-pricing] OpenRouter pricing fetch failed (timeout 60s): TimeoutError: The operation was aborted due to timeout
09:04:56 [model-pricing] LiteLLM pricing fetch failed (timeout 60s): TimeoutError: The operation was aborted due to timeout
09:05:15 [plugins] bonjour: watchdog detected non-announced service; attempting re-advertise (gateway fqdn=MP15-MS05 (OpenClaw)._openclaw-gw._tcp.local. host=MP15-MS05.local. port=18789 state=unannounced)

stiff minnowBOT Apr 28, 2026, 4:08 PM

#

Nice, this is progress. Two things in that output are worth cleaning up, but I don’t think they’re the core channel hang:

bonjour is stuck probing/re-advertising mDNS.
model pricing fetches are timing out after 60s.

Let’s remove that noise first:

openclaw config set models.pricing.enabled false --strict-json
openclaw plugins disable bonjour

Then stop the foreground gateway with Ctrl+C and retry with trace:

set OPENCLAW_GATEWAY_STARTUP_TRACE=1

#

set OPENCLAW_SKIP_CHANNELS=
openclaw gateway --port 18789

If it still doesn’t reach gateway ready, paste the new trace.

Also run this so we can see which channels are configured, without exposing tokens:

openclaw channels status

The earlier OPENCLAW_SKIP_CHANNELS=1 test proved the core Gateway is fine; we’re narrowing it to channel startup or one startup network sidecar.

broken rover Apr 28, 2026, 4:11 PM

#

09:10:45 [ws] unauthorized conn=f270b964-27a9-4ba0-b8b7-22ba91ef5a06 peer=127.0.0.1:50152->127.0.0.1:18789 remote=127.0.0.1 client=openclaw-control-ui webchat vcontrol-ui role=operator scopes=5 auth=token device=yes platform=Win32 instance=8452acd0-cb0b-4684-a6a5-b76433b0b34a host=127.0.0.1:18789 origin=http://127.0.0.1:18789 ua=Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/147.0.0.0 Safari/537.36 reason=device_token_mismatch
09:10:45 [ws] closed before connect conn=f270b964-27a9-4ba0-b8b7-22ba91ef5a06 peer=127.0.0.1:50152->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=http://127.0.0.1:18789 host=127.0.0.1:18789 ua=Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/147.0.0.0 Safari/537.36 code=1008 reason=unauthorized: device token mismatch (rotate/reissue device token)
09:11:00 [ws] unauthorized conn=2c5711f2-ba66-4065-b391-e57c1b047e2d peer=127.0.0.1:50153->127.0.0.1:18789 remote=127.0.0.1 client=openclaw-control-ui webchat vcontrol-ui role=operator scopes=5 auth=token device=yes platform=Win32 instance=8452acd0-cb0b-4684-a6a5-b76433b0b34a host=127.0.0.1:18789 origin=http://127.0.0.1:18789 ua=Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/147.0.0.0 Safari/537.36 reason=device_token_mismatch
09:11:00 [ws] closed before connect conn=2c5711f2-ba66-4065-b391-e57c1b047e2d peer=127.0.0.1:50153->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=http://127.0.0.1:18789 host=127.0.0.1:18789 ua=Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/147.0.0.0 Safari/537.36 code=1008 reason=unauthorized: device token mismatch (rotate/reissue device token)

stiff minnowBOT Apr 28, 2026, 4:11 PM

#

That device_token_mismatch is a separate Dashboard/browser auth issue, not the channel startup hang.

It means the Control UI has a stale cached device token. Since Gateway is at least accepting WebSocket attempts now, fix the browser side like this:

Open the dashboard in an incognito/private window, or clear site data for:

http://127.0.0.1:18789

Reconnect using the current Gateway token/password from the dashboard access panel / config. Don’t paste the token here.

If CLI commands also start showing device token mismatch, then rotate/re-approve via:

openclaw devices list

#

openclaw devices rotate --device <deviceId> --role operator

But first: did the gateway reach ready after disabling pricing + bonjour, or is it still stuck? Please paste the startup trace lines around starting channels and sidecars... again. That’s the actual boot problem; these WS lines are just the browser repeatedly trying with an old token.

Docs: https://docs.openclaw.ai/gateway/troubleshooting and https://docs.openclaw.ai/cli/devices

broken rover Apr 28, 2026, 4:15 PM

#

C:\Users\Eagle3dStreaming>openclaw config set models.pricing.enabled false --strict-json

🦞 OpenClaw 2026.4.26 (be8c246) — Finally, a use for that always-on Mac Mini under your desk.

Error: Config validation failed: models: Unrecognized key: "pricing"

C:\Users\Eagle3dStreaming>openclaw plugins disable bonjour

🦞 OpenClaw 2026.4.26 (be8c246) — I can run local, remote, or purely on vibes—results may vary with DNS.

Config overwrite: C:\Users\Eagle3dStreaming.openclaw\openclaw.json (sha256 affc56c166647bfca6ff0c4fe73d92999c99efa21e9469244d577851181ed251 -> d0d1f551106d219697246596bcc2a45b9d30cc9dc12cf0fe70017b245894e23d, backup=C:\Users\Eagle3dStreaming.openclaw\openclaw.json.bak)
Disabled plugin "bonjour". Restart the gateway to apply.

C:\Users\Eagle3dStreaming>

stiff minnowBOT Apr 28, 2026, 4:16 PM

#

Yep — that one is on me. models.pricing.enabled is on main/docs ahead of the published 2026.4.26; it’s not valid in 4.26, so ignore that. bonjour disabled correctly though.

Next steps:

set OPENCLAW_GATEWAY_STARTUP_TRACE=1
set OPENCLAW_SKIP_CHANNELS=
openclaw gateway --port 18789

Paste the starting channels and sidecars... section again after disabling bonjour.

Also run this to list configured channels without secrets:

node -e "const c=require(process.env.USERPROFILE+'/.openclaw/openclaw.json'); for (const [k,v] of Object.entries(c.channels||{})) console.log(k, 'enabled=', v&&v.enabled)"

#

Then we’ll disable/re-enable the actual configured channel(s), instead of guessing. The pricing timeout is annoying but not the main fix on 2026.4.26.

broken rover Apr 28, 2026, 4:18 PM

#

09:17:25 [gateway] log file: C:\Users\EAGLE3~1\AppData\Local\Temp\openclaw\openclaw-2026-04-28.log
09:17:25 [gateway] startup trace: post-attach.log 170.5ms total=24564.7ms eventLoopMax=0.0ms
09:17:25 [gateway] startup trace: post-attach.update-check 4.0ms total=24574.1ms eventLoopMax=0.0ms
09:17:25 [gateway] starting channels and sidecars...
09:17:25 [gateway] startup trace: sidecars.session-locks 9.6ms total=24595.5ms eventLoopMax=0.0ms
09:17:25 [gateway] startup trace: sidecars.gmail-watch 0.2ms total=24600.6ms eventLoopMax=0.0ms
09:17:25 [gateway] startup trace: sidecars.gmail-model 0.2ms total=24605.0ms eventLoopMax=0.0ms
09:17:25 [gateway] startup trace: sidecars.internal-hooks 0.2ms total=24609.7ms eventLoopMax=0.0ms
09:17:29 [ws] unauthorized conn=8f43d6e9-71dd-47a3-b583-7ef6ae3ee78c peer=127.0.0.1:50188->127.0.0.1:18789 remote=127.0.0.1 client=openclaw-control-ui webchat vcontrol-ui role=operator scopes=5 auth=token device=yes platform=Win32 instance=8452acd0-cb0b-4684-a6a5-b76433b0b34a host=127.0.0.1:18789 origin=http://127.0.0.1:18789 ua=Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/147.0.0.0 Safari/537.36 reason=device_token_mismatch
09:18:22 [ws] closed before connect conn=8f43d6e9-71dd-47a3-b583-7ef6ae3ee78c peer=127.0.0.1:50188->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=http://127.0.0.1:18789 host=127.0.0.1:18789 ua=Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/147.0.0.0 Safari/537.36 code=4008 reason=connect failed

stiff minnowBOT Apr 28, 2026, 4:19 PM

#

This confirms it’s still blocking inside sidecars.channels — there’s no sidecars.channels ... completion line.

Please run/paste this one; it’s the missing piece:

node -e "const c=require(process.env.USERPROFILE+'/.openclaw/openclaw.json'); for (const [k,v] of Object.entries(c.channels||{})) console.log(k, 'enabled=', v&&v.enabled)"

Then quickest recovery is: disable the enabled channel(s), restart, and re-enable one at a time.

Example:

openclaw config set channels.telegram.enabled false --strict-json
openclaw config set channels.whatsapp.enabled false --strict-json
openclaw gateway restart

#

Also close the Dashboard tab for now — the device_token_mismatch spam is separate noise from a stale browser token. Once Gateway is healthy, clear site data for 127.0.0.1:18789 or use incognito.

broken rover Apr 28, 2026, 4:20 PM

#

i just ran this command, and now what?

node -e "const c=require(process.env.USERPROFILE+'/.openclaw/openclaw.json'); for (const [k,v] of Object.entries(c.channels||{})) console.log(k, 'enabled=', v&&v.enabled)"

stiff minnowBOT Apr 28, 2026, 4:20 PM

#

What did it print? Paste the output.

If it printed nothing, that means there are no channels entries in config, and we’ll need to look at plugins instead:

openclaw plugins list --enabled

If it printed channel names, disable the ones showing enabled= true, then restart. Example:

openclaw config set channels.telegram.enabled false --strict-json
openclaw gateway restart

Replace telegram with whatever the command printed.

broken rover Apr 28, 2026, 4:20 PM

#

C:\Users\Eagle3dStreaming>node -e "const c=require(process.env.USERPROFILE+'/.openclaw/openclaw.json'); for (const [k,v] of Object.entries(c.channels||{})) console.log(k, 'enabled=', v&&v.enabled)"
telegram enabled= false
whatsapp enabled= false

C:\Users\Eagle3dStreaming>

stiff minnowBOT Apr 28, 2026, 4:21 PM

#

Got it — Telegram and WhatsApp are already disabled at the channel config layer, so the hang is probably a bundled channel plugin startup/listing path, not an enabled account.

Stop the foreground gateway with Ctrl+C, then hard-disable those plugins too:

openclaw plugins disable telegram
openclaw plugins disable whatsapp
openclaw gateway restart

Then check:

openclaw gateway status
openclaw status

#

If it still hangs, paste:

openclaw plugins list --enabled --verbose

This is a bit unexpected with both channel flags false, so if plugin-disable fixes it, that’s useful bug-report material.

broken rover Apr 28, 2026, 4:28 PM

#

its just sittign there with 'probing gateway'

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: running (last run 267009, last run time 4/28/2026 9:22:23 AM)
Warm-up: launch agents can take a few seconds. Try again shortly.
Connectivity probe: failed
Probe target: ws://127.0.0.1:18789
timeout
Capability: unknown

Port 18789 is already in use.

pid 5200: "C:\Program Files\nodejs\node.exe" C:\Users\Eagle3dStreaming\AppData\Roaming\npm\node_modules\openclaw\dist\index.js gateway --port 18789 (127.0.0.1:18789)
Gateway already running locally. Stop it (openclaw gateway stop) or use a different port.
Listening: 127.0.0.1:18789
Troubles: run openclaw status
Troubleshooting: https://docs.openclaw.ai/troubleshooting

C:\Users\Eagle3dStreaming>openclaw status

🦞 OpenClaw 2026.4.26 (be8c246) — WhatsApp automation without the "please accept our new privacy policy".

|
• Probing gateway…..

stiff minnowBOT Apr 28, 2026, 4:29 PM

#

Yep — openclaw status is now waiting on a Gateway process that is listening on the port but not answering RPC. Don’t wait on that; hit Ctrl+C.

Then reset the stuck process:

openclaw gateway stop

If it still says port 18789 is in use by PID 5200, kill just that stuck Gateway process:

taskkill /PID 5200 /T /F

Then run the foreground trace again so we can see what it does after disabling bonjour, telegram, and whatsapp:

#

set OPENCLAW_GATEWAY_STARTUP_TRACE=1
openclaw gateway --port 18789

If it hangs again, paste the lines after:

starting channels and sidecars...

Also paste this local inventory:

openclaw plugins list --enabled --verbose

#

We’re past the “is Gateway installed?” stage — now it’s a stuck sidecar/plugin during startup.

broken rover Apr 28, 2026, 4:30 PM

#

C:\Users\Eagle3dStreaming>openclaw gateway stop

🦞 OpenClaw 2026.4.26 (be8c246) — I'll butter your workflow like a lobster roll: messy, delicious, effective.

Stopped Scheduled Task: OpenClaw Gateway

C:\Users\Eagle3dStreaming>

stiff minnowBOT Apr 28, 2026, 4:30 PM

#

Good. Now verify the port is actually free:

openclaw gateway status

If it says port 18789 is still in use, paste that and we’ll kill the remaining PID.

If it’s free, run:

set OPENCLAW_GATEWAY_STARTUP_TRACE=1
openclaw gateway --port 18789

Then paste the lines after starting channels and sidecars....

#

Also useful, before/after the trace:

openclaw plugins list --enabled --verbose

broken rover Apr 28, 2026, 4:31 PM

#

C:\Users\Eagle3dStreaming>openclaw gateway status

🦞 OpenClaw 2026.4.26 (be8c246) — Your .zshrc wishes it could do what I do.

|
o
Service: Scheduled Task (registered)
File logs: C:\Users\EAGLE3~1\AppData\Local\Temp\openclaw\openclaw-2026-04-28.log
Command: C:\Program Files\nodejs\node.exe C:\Users\Eagle3dStreaming\AppData\Roaming\npm\node_modules\openclaw\dist\index.js gateway --port 18789
Service file: ~.openclaw\gateway.cmd
Service env: OPENCLAW_GATEWAY_PORT=18789

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: unknown (state Ready, last run -1073741510, last run time 4/28/2026 9:22:23 AM, Task status is locale-dependent and no numeric Last Run Result was available.)
Connectivity probe: failed
Probe target: ws://127.0.0.1:18789
connect ECONNREFUSED 127.0.0.1:18789
Capability: unknown

Troubles: run openclaw status
Troubleshooting: https://docs.openclaw.ai/troubleshooting

C:\Users\Eagle3dStreaming>

stiff minnowBOT Apr 28, 2026, 4:31 PM

#

Good — port is free now. The -1073741510 just means the previous foreground/Scheduled Task was interrupted; not scary.

Now run:

set OPENCLAW_GATEWAY_STARTUP_TRACE=1
openclaw gateway --port 18789

Wait until either it prints ready or it sits for ~60s, then paste the section after:

starting channels and sidecars...

If you want the plugin inventory first:

#

openclaw plugins list --enabled --verbose

broken rover Apr 28, 2026, 4:32 PM

#

09:32:00 [gateway] startup trace: runtime.config 4.7ms total=23947.4ms eventLoopMax=0.0ms
09:32:00 [gateway] startup trace: control-ui.root 6.2ms total=23961.4ms eventLoopMax=0.0ms
09:32:00 [gateway] startup trace: tls.runtime 0.3ms total=23967.1ms eventLoopMax=0.0ms
09:32:00 [gateway] starting HTTP server...
09:32:00 [canvas] host mounted at http://127.0.0.1:18789/__openclaw__/canvas/ (root C:\Users\Eagle3dStreaming.openclaw\canvas)
09:32:00 [gateway] startup trace: runtime.state 15.5ms total=23993.2ms eventLoopMax=0.0ms
09:32:00 [gateway] startup trace: runtime.early 438.5ms total=24441.7ms eventLoopMax=0.0ms
09:32:00 [health-monitor] started (interval: 300s, startup-grace: 60s, channel-connect-grace: 120s)
09:32:01 [gateway] startup trace: http.bound 367.7ms total=24809.5ms eventLoopMax=0.0ms
09:32:01 [gateway] startup trace: post-attach.update-sentinel 132.2ms total=24946.9ms eventLoopMax=0.0ms
09:32:01 [gateway] agent model: anthropic/claude-sonnet-4-6
09:32:01 [gateway] http server listening (6 plugins: acpx, browser, device-pair, memory-core, phone-control, talk-voice; 27.6s)
09:32:01 [gateway] log file: C:\Users\EAGLE3~1\AppData\Local\Temp\openclaw\openclaw-2026-04-28.log
09:32:01 [gateway] startup trace: post-attach.log 170.5ms total=25122.0ms eventLoopMax=0.0ms
09:32:01 [gateway] startup trace: post-attach.update-check 3.7ms total=25130.1ms eventLoopMax=0.0ms
09:32:01 [gateway] starting channels and sidecars...
09:32:01 [gateway] startup trace: sidecars.session-locks 13.7ms total=25153.6ms eventLoopMax=0.0ms
09:32:01 [gateway] startup trace: sidecars.gmail-watch 0.2ms total=25159.1ms eventLoopMax=0.0ms
09:32:01 [gateway] startup trace: sidecars.gmail-model 0.3ms total=25166.1ms eventLoopMax=0.0ms
09:32:01 [gateway] startup trace: sidecars.internal-hooks 0.2ms total=25171.1ms eventLoopMax=0.0ms

stiff minnowBOT Apr 28, 2026, 4:33 PM

#

Still the same spot: sidecars.channels. Since telegram/whatsapp are false and bonjour is gone, I’d switch to recovery + bug report mode.

For now, make Gateway usable by persistently skipping channel startup:

echo OPENCLAW_SKIP_CHANNELS=1>> "%USERPROFILE%\.openclaw\.env"
openclaw gateway restart
openclaw gateway status

That should get you back to gateway ready like the earlier test. It disables chat channel providers, but keeps the Gateway/Dashboard usable.

Then please grab these for a GitHub issue:

openclaw plugins list --enabled --verbose

#

openclaw status --all

Issue link: https://github.com/openclaw/openclaw/issues

The key repro detail is: OpenClaw 2026.4.26 on Windows hangs in sidecars.channels; OPENCLAW_SKIP_CHANNELS=1 reaches ready immediately; configured telegram and whatsapp are both enabled=false.

GitHub

Issues · openclaw/openclaw

Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞 - Issues · openclaw/openclaw

broken rover Apr 28, 2026, 4:38 PM

#

i still can't open the web ui

09:34:55 [gateway] starting...
09:35:22 [gateway] starting HTTP server...
09:35:22 [canvas] host mounted at http://127.0.0.1:18789/__openclaw__/canvas/ (root C:\Users\Eagle3dStreaming.openclaw\canvas)
09:35:23 [health-monitor] started (interval: 300s, startup-grace: 60s, channel-connect-grace: 120s)
09:35:23 [gateway] agent model: anthropic/claude-sonnet-4-6
09:35:23 [gateway] http server listening (6 plugins: acpx, browser, device-pair, memory-core, phone-control, talk-voice; 28.3s)
09:35:23 [gateway] log file: C:\Users\EAGLE3~1\AppData\Local\Temp\openclaw\openclaw-2026-04-28.log
09:35:23 [gateway] starting channels and sidecars...
09:35:27 [ws] unauthorized conn=02361def-73f9-4dff-8f01-67663dc15bef peer=127.0.0.1:50271->127.0.0.1:18789 remote=127.0.0.1 client=openclaw-control-ui webchat vcontrol-ui role=operator scopes=5 auth=token device=yes platform=Win32 instance=8452acd0-cb0b-4684-a6a5-b76433b0b34a host=127.0.0.1:18789 origin=http://127.0.0.1:18789 ua=Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/147.0.0.0 Safari/537.36 reason=device_token_mismatch
09:36:21 [ws] closed before connect conn=02361def-73f9-4dff-8f01-67663dc15bef peer=127.0.0.1:50271->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=http://127.0.0.1:18789 host=127.0.0.1:18789 ua=Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/147.0.0.0 Safari/537.36 code=4008 reason=connect failed
09:36:21 [ws] closed before connect conn=ba05fe45-3d77-451c-bb0b-ac99b52db591 peer=127.0.0.1:50278->127.0.0.1:18789 remote=127.0.0.1 fwd=n/a origin=http://127.0.0.1:18789 host=127.0.0.1:18789 ua=Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko)

stiff minnowBOT Apr 28, 2026, 4:38 PM

#

The log shows it did not apply OPENCLAW_SKIP_CHANNELS=1 — if it had, you’d see:

[gateway/channels] skipping channel start ...
[gateway] ready

Do this recovery path:

Stop the stuck gateway:

openclaw gateway stop

Open the env file:

#

notepad "%USERPROFILE%\.openclaw\.env"

Add this line, save, close Notepad:

OPENCLAW_SKIP_CHANNELS=1

Restart:

openclaw gateway restart
openclaw gateway status

#

Once gateway status says the probe is OK / ready, open the UI with:

openclaw dashboard

And because your browser has device_token_mismatch, use an incognito window or clear site data for http://127.0.0.1:18789 first. That browser token issue is separate, but it will block the UI even after Gateway is healthy.

broken rover Apr 28, 2026, 4:45 PM

#

i added that line to the env file but now i'm jst stuck here

C:\Users\Eagle3dStreaming>openclaw gateway stop

🦞 OpenClaw 2026.4.26 (be8c246) — I autocomplete your thoughts—just slower and with more API calls.

Stopped Scheduled Task: OpenClaw Gateway

C:\Users\Eagle3dStreaming>notepad "%USERPROFILE%.openclaw.env"

C:\Users\Eagle3dStreaming>openclaw gateway restart

🦞 OpenClaw 2026.4.26 (be8c246) — I'm not saying your workflow is chaotic... I'm just bringing a linter and a helmet.

Restarted Scheduled Task: OpenClaw Gateway

C:\Users\Eagle3dStreaming>openclaw status

🦞 OpenClaw 2026.4.26 (be8c246) — Your terminal just grew claws—type something and let the bot pinch the busywork.

|
• Probing gateway…..

stiff minnowBOT Apr 28, 2026, 4:46 PM

#

Hit Ctrl+C — don’t wait on openclaw status right now.

Let’s verify whether the .env is actually being read:

type "%USERPROFILE%\.openclaw\.env"

Make sure it shows exactly:

OPENCLAW_SKIP_CHANNELS=1

Then bypass the Scheduled Task once, with the env set explicitly:

#

openclaw gateway stop
set OPENCLAW_SKIP_CHANNELS=1
set OPENCLAW_GATEWAY_STARTUP_TRACE=1
openclaw gateway --port 18789

If that prints skipping channel start... and then gateway ready, leave that window open and try:

openclaw dashboard

For the Web UI auth issue: use an incognito window or clear site data for 127.0.0.1:18789, because your logs showed device_token_mismatch from the browser.

broken rover Apr 28, 2026, 4:47 PM

#

C:\Users\Eagle3dStreaming>type "%USERPROFILE%.openclaw.env"
VITE_SUPABASE_URL=https://qohgakoopdrafaolcxsl.supabase.co
VITE_SUPABASE_ANON_KEY=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJzdXBhYmFzZSIsInJlZiI6InFvaGdha29vcGRyYWZhb2xjeHNsIiwicm9sZSI6ImFub24iLCJpYXQiOjE3NzY2MjQ4ODAsImV4cCI6MjA5MjIwMDg4MH0.ykjQeHrPrGY4XJAeGqXC4x8MiUXBx6G4dQDUZCZbktM
SUPABASE_SERVICE_ROLE_KEY=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJzdXBhYmFzZSIsInJlZiI6InFvaGdha29vcGRyYWZhb2xjeHNsIiwicm9sZSI6InNlcnZpY2Vfcm9sZSIsImlhdCI6MTc3NjYyNDg4MCwiZXhwIjoyMDkyMjAwODgwfQ.fcexchrmhst1s8v2zhRjvPo32gFQm4AuA8uEjIbksrI
DATABASE_URL=postgresql://postgres.qohgakoopdrafaolcxsl:mwpaaa5j!Qazxsw2@aws-1-us-east-1.pooler.supabase.com:5432/postgresOPENCLAW_SKIP_CHANNELS=
OPENCLAW_SKIP_CHANNELS=1

C:\Users\Eagle3dStreaming>

stiff minnowBOT Apr 28, 2026, 4:47 PM

#

Two important things:

You pasted live secrets publicly — especially SUPABASE_SERVICE_ROLE_KEY and DATABASE_URL. Assume they’re compromised. Rotate/revoke them in Supabase ASAP, and delete/edit that Discord message if you can.
Your .env is malformed here:

.../postgresOPENCLAW_SKIP_CHANNELS=
OPENCLAW_SKIP_CHANNELS=1

There’s no newline after DATABASE_URL, and there’s a blank OPENCLAW_SKIP_CHANNELS= entry.

Open it:

#

notepad "%USERPROFILE%\.openclaw\.env"

Fix the bottom so each variable is on its own line, and remove the empty one:

DATABASE_URL=postgresql://...
OPENCLAW_SKIP_CHANNELS=1

Save, then:

openclaw gateway stop
openclaw gateway restart

#

openclaw gateway status

If it still hangs, run foreground once:

set OPENCLAW_SKIP_CHANNELS=1
set OPENCLAW_GATEWAY_STARTUP_TRACE=1
openclaw gateway --port 18789

broken rover Apr 28, 2026, 4:49 PM

#

C:\Users\Eagle3dStreaming>set OPENCLAW_SKIP_CHANNELS=1

C:\Users\Eagle3dStreaming>set OPENCLAW_GATEWAY_STARTUP_TRACE=1

C:\Users\Eagle3dStreaming>openclaw gateway --port 18789

🦞 OpenClaw 2026.4.26 (be8c246) — Your terminal just grew claws—type something and let the bot pinch the busywork.

|
o
09:49:10 [gateway] startup trace: cli.server-import 6.8ms total=6.8ms
09:49:10 [gateway] loading configuration…
09:49:11 [gateway] startup trace: cli.config-load 187.1ms total=229.8ms
09:49:11 [gateway] startup trace: cli.config-snapshot 184.2ms total=418.6ms
09:49:11 [gateway] resolving authentication…
09:49:11 [gateway] startup trace: cli.auth-resolve 0.7ms total=432.2ms
09:49:11 [gateway] starting...
09:49:11 [gateway] startup trace: cli.gateway-loop 10.8ms total=443.0ms
09:49:16 Gateway failed to start: gateway already running (pid 6948); lock timeout after 5000ms
If the gateway is supervised, stop it with: openclaw gateway stop
09:49:16 Port 18789 is already in use.
09:49:16 - pid 6948: "C:\Program Files\nodejs\node.exe" C:\Users\Eagle3dStreaming\AppData\Roaming\npm\node_modules\openclaw\dist\index.js gateway --port 18789 (127.0.0.1:18789)
09:49:16 - Gateway already running locally. Stop it (openclaw gateway stop) or use a different port.
09:49:16 Gateway service appears registered. Stop it first.
09:49:16 Tip: openclaw gateway stop
09:49:16 Or: schtasks /End /TN "OpenClaw Gateway"

C:\Users\Eagle3dStreaming>

stiff minnowBOT Apr 28, 2026, 4:49 PM

#

You still have the Scheduled Task gateway running as PID 6948, so the foreground one can’t start.

Run:

openclaw gateway stop

Then verify/kill if needed:

tasklist /FI "PID eq 6948"

If it still shows node.exe, run:

#

taskkill /PID 6948 /T /F

Then start foreground again:

set OPENCLAW_SKIP_CHANNELS=1
set OPENCLAW_GATEWAY_STARTUP_TRACE=1
openclaw gateway --port 18789

You’re looking for this line:

[gateway/channels] skipping channel start ...

#

[gateway] ready

And seriously: rotate those Supabase keys — the service role key is basically admin access.

broken rover Apr 28, 2026, 4:51 PM

#

09:51:28 [gateway] startup trace: sidecars.gmail-model 0.3ms total=32331.2ms eventLoopMax=0.0ms
09:51:28 [gateway] startup trace: sidecars.internal-hooks 0.2ms total=32335.6ms eventLoopMax=0.0ms
09:51:29 [gateway/channels] skipping channel start (OPENCLAW_SKIP_CHANNELS=1 or OPENCLAW_SKIP_PROVIDERS=1)
09:51:29 [gateway] startup trace: sidecars.channels 4.4ms total=32344.5ms eventLoopMax=0.0ms
09:51:29 [plugins] embedded acpx runtime backend registered (cwd: C:\Users\Eagle3dStreaming.openclaw\workspace)
09:51:33 [browser/server] Browser control listening on http://127.0.0.1:18791/ (auth=token)
09:51:33 [gateway] startup trace: sidecars.plugin-services 4340.5ms total=36692.4ms eventLoopMax=4315.9ms
09:51:33 [gateway] startup trace: sidecars.memory 0.3ms total=36697.2ms eventLoopMax=0.0ms
09:51:33 [gateway] startup trace: sidecars.restart-sentinel 14.2ms total=36715.8ms eventLoopMax=0.0ms
09:51:33 [gateway] startup trace: sidecars.subagent-recovery 5.7ms total=36726.3ms eventLoopMax=0.0ms
09:51:33 [gateway] startup trace: sidecars.main-session-recovery 64.8ms total=36806.1ms eventLoopMax=0.0ms
09:51:33 [gateway] startup trace: sidecars.total 4499.6ms total=36811.1ms eventLoopMax=0.0ms
09:51:33 [gateway] startup trace: sidecars.ready 4.5ms total=36815.5ms eventLoopMax=0.0ms
09:51:33 [gateway] ready
09:51:33 [gateway] startup trace: runtime.post-attach 5627.2ms total=36827.7ms eventLoopMax=0.0ms
09:51:33 [gateway] startup trace: ready 5.0ms total=36832.7ms eventLoopMax=0.0ms
09:51:33 [heartbeat] started

stiff minnowBOT Apr 28, 2026, 4:51 PM

#

Great — Gateway is healthy now. 🎉 The confirmed workaround is OPENCLAW_SKIP_CHANNELS=1; the broken part is channel startup in 2026.4.26.

Leave that foreground window open for now. In a new cmd window, run:

openclaw gateway status
openclaw dashboard

For the Web UI, because the logs showed device_token_mismatch, open it in an incognito/private window or clear site data for:

http://127.0.0.1:18789

Then reconnect.

#

After you’re back in, please file an issue with this summary:

OpenClaw 2026.4.26 on Windows 10 hangs at sidecars.channels.
OPENCLAW_SKIP_CHANNELS=1 reaches gateway ready.
channels.telegram.enabled=false and channels.whatsapp.enabled=false.
bonjour plugin disabled; still hangs without OPENCLAW_SKIP_CHANNELS.

Also rotate those Supabase secrets you pasted. That’s urgent.

broken rover Apr 28, 2026, 4:53 PM

#

C:\Users\Eagle3dStreaming>openclaw gateway status

🦞 OpenClaw 2026.4.26 (be8c246) — Pairing codes exist because even bots believe in consent—and good security hygiene.

|
• Checking gateway status...enclaw dashboard
o
Service: Scheduled Task (registered)
File logs: C:\Users\EAGLE3~1\AppData\Local\Temp\openclaw\openclaw-2026-04-28.log
Command: C:\Program Files\nodejs\node.exe C:\Users\Eagle3dStreaming\AppData\Roaming\npm\node_modules\openclaw\dist\index.js gateway --port 18789
Service file: ~.openclaw\gateway.cmd
Service env: OPENCLAW_GATEWAY_PORT=18789

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: running (pid 2452, state Ready, last run 267014, last run time 4/28/2026 9:41:01 AM, Verified gateway listener detected on port 18789 even though schtasks did not report a running task.)
Warm-up: launch agents can take a few seconds. Try again shortly.
Connectivity probe: failed
Probe target: ws://127.0.0.1:18789
timeout
Capability: unknown

Port 18789 is already in use.

pid 2452: "node" "C:\Users\Eagle3dStreaming\AppData\Roaming\npm\node_modules\openclaw\openclaw.mjs" gateway --port 18789 (127.0.0.1:18789)
Gateway already running locally. Stop it (openclaw gateway stop) or use a different port.
Listening: 127.0.0.1:18789
Troubles: run openclaw status
Troubleshooting: https://docs.openclaw.ai/troubleshooting

C:\Users\Eagle3dStreaming>

stiff minnowBOT Apr 28, 2026, 4:53 PM

#

That output means the port is occupied by the foreground Gateway you started — that part is expected. But the probe timing out means the client/UI auth cache is probably still poisoned by the earlier device_token_mismatch.

Do this in a new cmd window:

ren "%USERPROFILE%\.openclaw\identity\device-auth.json" "device-auth.json.bak"

If that says file not found, that’s okay.

Then try:

openclaw gateway status
openclaw dashboard

#

For the browser UI: use an incognito/private window, or clear site data for 127.0.0.1:18789, because the browser has a stale device token.

Also: once you’re stable, fix/rotate the Supabase secrets you pasted publicly. The service role key should be treated as compromised.

broken rover Apr 28, 2026, 4:54 PM

#

C:\Users\Eagle3dStreaming>ren "%USERPROFILE%.openclaw\identity\device-auth.json" "device-auth.json.bak"

C:\Users\Eagle3dStreaming>

stiff minnowBOT Apr 28, 2026, 4:54 PM

#

Good — that cleared the CLI cached device token.

Now, in that same new cmd window, try:

openclaw gateway status
openclaw dashboard --no-open

Open the dashboard URL in an incognito/private window first. If you use the existing browser tab, it may still have the stale Control UI device token cached.

If it still fails in the browser, clear site data for:

http://127.0.0.1:18789

#

Then reload the dashboard.

And please don’t paste .env again — it had live Supabase secrets. Rotate those keys when you’re out of the fire.

broken rover Apr 28, 2026, 4:55 PM

#

:\Users\Eagle3dStreaming>openclaw gateway status

🦞 OpenClaw 2026.4.26 (be8c246) — Self-hosted, self-updating, self-aware (just kidding... unless?).

|
o
Service: Scheduled Task (registered)
File logs: C:\Users\EAGLE3~1\AppData\Local\Temp\openclaw\openclaw-2026-04-28.log
Command: C:\Program Files\nodejs\node.exe C:\Users\Eagle3dStreaming\AppData\Roaming\npm\node_modules\openclaw\dist\index.js gateway --port 18789
Service file: ~.openclaw\gateway.cmd
Service env: OPENCLAW_GATEWAY_PORT=18789

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: running (pid 2452, state Ready, last run 267014, last run time 4/28/2026 9:41:01 AM, Verified gateway listener detected on port 18789 even though schtasks did not report a running task.)
Warm-up: launch agents can take a few seconds. Try again shortly.
Connectivity probe: failed
Probe target: ws://127.0.0.1:18789
timeout
Capability: unknown

Port 18789 is already in use.

pid 2452: "node" "C:\Users\Eagle3dStreaming\AppData\Roaming\npm\node_modules\openclaw\openclaw.mjs" gateway --port 18789 (127.0.0.1:18789)
Gateway already running locally. Stop it (openclaw gateway stop) or use a different port.
Listening: 127.0.0.1:18789
Troubles: run openclaw status
Troubleshooting: https://docs.openclaw.ai/troubleshooting

C:\Users\Eagle3dStreaming>

stiff minnowBOT Apr 28, 2026, 4:55 PM

#

Okay — Gateway is listening, but RPC probe is timing out. Let’s split HTTP vs WebSocket.

Run these:

curl http://127.0.0.1:18789/healthz
curl http://127.0.0.1:18789/readyz

Then try opening the UI directly in an incognito/private window:

http://127.0.0.1:18789/

If the page loads but says disconnected/unauthorized, clear site data for 127.0.0.1:18789 — your earlier logs showed stale browser device auth.

#openclaw stuck on booting gateway