#Openclaw Codex 5.5
1 messages Β· Page 1 of 1 (latest)
Ja β wenn du mit .25 2026.4.25 meinst, ist der richtige Weg fΓΌr ChatGPT/Codex-Subscription:
openclaw models auth login --provider openai-codex
openclaw config set agents.defaults.model.primary openai-codex/gpt-5.5
openclaw models status
FΓΌr VPS/headless:
openclaw models auth login --provider openai-codex --device-code
Wichtig: openai-codex/gpt-5.5 ist der Codex/OAuth/Subscription-Pfad. openai/gpt-5.5 wΓ€re der direkte OpenAI-API-Key-Pfad bzw. mit extra Codex-Harness-Konfig. OpenAI rollt GPTβ5.5 gerade fΓΌr ChatGPT/Codex aus, aber API-VerfΓΌgbarkeit kann separat sein. (help.openai.com)
Zum Testen in einer bestehenden Session kannst du auch einfach senden:
/model openai-codex/gpt-5.5
Docs: https://docs.openclaw.ai/providers/openai π¦
Ich bekomme jetzt nachdem ich meiner main session gesagt habe sie soll einrichten, dass ich codex 5.5 verwenden kann, keine reaktion mehr, und ein history failed: Error: gateway request timeout for chat.history.
Exec approval is required, but no interactive approval client is currently available.
Approve it from the Web UI or terminal UI, or enable a native chat approval client such as Discord. If those accounts already know your owner ID via allowFrom or owner config, OpenClaw can often infer approvers automatically. Then retry the command. You can usually leave execApprovals.approvers unset when owner config already identifies the approvers.
Automatic session resume failed, so sending the status directly.
Exec denied (gateway id=6bd837a0-1d1c-4b99-8565-6375841f74cf, approval-timeout (allowlist-miss)): if [ -d ~/openclaw-ref/.git ]; then cd ~/openclaw-ref && git pull --ff-only --quiet && echo pulled; else echo missing; fi
Das klingt eher nach Gateway/UI oder Session hΓ€ngt, nicht zwingend nach βCodex 5.5 falschβ.
Mach bitte zuerst den kurzen Recovery-Loop:
-
In der betroffenen Session:
/stopoder in Control UI den Stop Button. Das bricht aktive Runs ab. (docs.openclaw.ai)
-
Wenn weiter
chat.historytimeout kommt: Gateway neu starten:openclaw gateway restart -
Danach bitte posten:
openclaw --version
openclaw status --all
openclaw models status
openclaw gateway status
Wichtig: Wenn deine Main-Session ein Modell gesetzt hat, das nicht in agents.defaults.models erlaubt ist oder fΓΌr deinen Auth nicht erreichbar ist, kann OpenClaw vor der normalen Antwort abbrechen β das fΓΌhlt sich dann wie βkeine Reaktionβ an. Die Docs empfehlen dann: Modell zur Allowlist hinzufΓΌgen, Allowlist entfernen, oder eins aus /model list wΓ€hlen. (docs.openclaw.ai)
Ich wΓΌrde nicht blind einen geratenen βCodex 5.5β Modellnamen hardcoden. Nimm das, was bei dir in /model list bzw. openclaw models list wirklich auftaucht; OpenAI/OpenClaw Modellnamen Γ€ndern sich schnell, und die OpenAI-Modellseite ist dafΓΌr die Quelle der Wahrheit. (platform.openai.com)
Kurzes Meta-Update: Mein Versuch, den lokalen OpenClaw-Ref-Checkout zu aktualisieren, ist wegen Approval-Timeout nicht gelaufen. Ich hab deshalb die aktuellen Docs online gegengecheckt.
FΓΌr dich Γ€ndert sich nichts: bitte mit
openclaw --version
openclaw status --all
openclaw models status
openclaw gateway status
weitermachen. Besonders models status sollte zeigen, ob der Codex/OpenAI-Code Provider sauber authβd ist oder ob die Session auf ein nicht erlaubtes/nicht erreichbares Modell zeigt.
Ich muss es in mehreren Texten schicken. Es ist zu viel. Einen moment
Kein Stress β schickβs ruhig in mehreren Teilen.
Am besten jeweils mit Labels, z.B.:
[1/4] openclaw --version
...
[2/4] openclaw status --all
...
[3/4] openclaw models status
...
[4/4] openclaw gateway status
...
Bitte Tokens/API keys vorher rausnehmen, falls irgendwo welche auftauchen.
1/10
major@localhost-live:~$ openclaw --version
openclaw status --all
openclaw models status
openclaw gateway status
OpenClaw 2026.4.25 (aa36ee6)
π¦ OpenClaw 2026.4.25 (aa36ee6) β More integrations than your therapist's intake form.
β
OpenClaw status --all
Overview
ββββββββββββββββββββββ¬ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β Item β Value β
ββββββββββββββββββββββΌββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β Version β 2026.4.25 β
β OS β linux 6.18.8-200.fc43.x86_64 (x64) β
Angekommen: OpenClaw 2026.4.25 auf Fedora/Linux.
Schick weiter β ich warte, bis alle 10 Teile da sind, bevor ichβs zusammen bewerte.
2/10
β Node β 22.22.0 β
β Config β ~/.openclaw/openclaw.json β
β Dashboard β http://127.0.0.1:18789/ β
β Tailscale exposure β off β
β Channel β stable (config) β
β Update β pnpm Β· npm update 2026.4.26 β
β Gateway β local Β· ws://127.0.0.1:18789 (local loopback) Β· unreachable (timeout) β
β Security β Run: openclaw security audit --deep β
β Gateway self β unknown β
Got it β das ist schon ein wichtiger Hinweis: Gateway unreachable / timeout auf ws://127.0.0.1:18789.
Bitte weiter schicken; ich sammle noch die restlichen Teile, bevor ich dir die sauberste Reihenfolge gebe.
3/10
β Gateway service β systemd installed Β· enabled Β· running (pid 3057569, state active) β
β Node service β systemd not installed β
β Agents β 5 total Β· 2 bootstrapping Β· 1 active Β· 13 sessions β
β Secrets β none
ββββββββββββββββββββββ΄ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
Danke β Service lΓ€uft zwar, aber RPC/WebSocket antwortet nicht. Das riecht nach Gateway-Prozess hΓ€ngt / blockiert / Health probe timed out.
Schick bitte weiter bis 10/10.
4/10
Channels
ββββββββββββ¬ββββββββββ¬βββββββββ¬βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β Channel β Enabled β State β Detail β
ββββββββββββΌββββββββββΌβββββββββΌβββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
ββββββββββββ΄ββββββββββ΄βββββββββ΄βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
Agents
βββββββββββββββββββββββββββββ¬βββββββββββββββββ¬βββββββββββ¬βββββββββββ¬ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β Agent β Bootstrap file β Sessions β Active β Store β
βββββββββββββββββββββββββββββΌβββββββββββββββββΌβββββββββββΌβββββββββββΌββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β main β ABSENT β 4 β 3m ago β ~/.openclaw/agents/main/sessions/sessions.json β β
π weiter, bitte. The main agent war vor 3 Minuten aktiv, passt also zu βich hab ihm gerade gesagtβ¦β β aber ich warte noch auf Logs/Model-Status.
5/10
β bewerbungen (bewerbungen) β PRESENT β 2 β 13d ago β ~/.openclaw/agents/bewerbungen/sessions/sessions.json β
β pokemon (pokemon) β ABSENT β 3 β 17d ago β ~/.openclaw/agents/pokemon/sessions/sessions.json β
β trading (trading) β ABSENT β 2 β 20m ago β ~/.openclaw/agents/trading/sessions/sessions.json β
β voidbloom (voidbloom) β PRESENT β 2 β 22d ago β ~/.openclaw/agents/voidbloom/sessions/sessions.json β
βββββββββββββββββββββββββββββ΄βββββββββββββββββ΄βββββββββββ΄βββββββββββ΄ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
Diagnosis (read-only)
Gateway connection details:
Gateway target: ws://127.0.0.1:18789
Source: local loopback
Config: /home/major/.openclaw/openclaw.json
Bind: loopback
β Config: /home/major/.openclaw/openclaw.json
β Secret diagnostics (0)
β Restart sentinel: none
! Port 18789
Port 18789 is already in use.
- pid 3057569 major: openclaw-gateway ([::1]:18789)
- Gateway already running locally. Stop it (openclaw gateway stop) or use a different port.
! Tailscale exposure: off Β· daemon unknown
β Skills: 17 eligible Β· 0 missing Β· /home/major/.openclaw/workspace
β Plugin compatibility (none)
! Channel issues skipped (gateway unreachable)
Gateway health:
timeout
Pasteable debug report. Auth tokens redacted.
Troubleshooting: https://docs.openclaw.ai/troubleshooting
β
π¦ OpenClaw 2026.4.25 (aa36ee6) β I run on caffeine, JSON5, and the audacity of "it worked on my machine."
6/10
Config : ~/.openclaw/openclaw.json
Agent dir : ~/.openclaw/agents/main/agent
Default : openai-codex/gpt-5.5
Fallbacks (2) : openai-codex/gpt-5.4,
[29.04.26 23:11] Tobias: anthropic/claude-sonnet-4-6
Image model : -
Image fallbacks (0): -
Aliases (2) : sonnet -> anthropic/claude-sonnet-4-6, opus -> anthropic/claude-opus-4-7
Configured models (8): openai-codex/gpt-5.4, anthropic/claude-sonnet-4-6, anthropic/claude-opus-4-6, anthropic/claude-opus-4-5, anthropic/claude-sonnet-4-5, anthropic/claude-haiku-4-5, openai-codex/gpt-5.5, anthropic/claude-opus-4-7
Auth overview
Auth store : ~/.openclaw/agents/main/agent/auth-profiles.json
Shell env : off
Providers w/ OAuth/tokens (3): anthropic (2), claude-cli (2), openai-codex (2)
- anthropic effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=2 (oauth=1, token=1, api_key=0) | anthropic:default=token:sk-ant-o...DjKtywAA, anthropic:claude-cli=OAuth
- claude-cli effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=2 (oauth=1, token=1, api_key=0) | anthropic:default=token:sk-ant-o...DjKtywAA, anthropic:claude-cli=OAuth | synthetic=plugin-owned | source=Claude CLI native auth
- geminicli2api effective=models.json:op...26 | models.json=op...26 | source=models.json: ~/.openclaw/agents/main/agent/models.json
- google effective=env:AIzaSyB3...10lXjJI8 | env=AIzaSyB3...10lXjJI8 | source=env: GEMINI_API_KEY
- openai-codex effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=2 (oauth=2, token=0, api_key=0) | openai-codex:default=OAuth, openai-codex:majorkarpati@gmail.com=OAuth (majorkarpati@gmail.com)
7/10
OAuth/token status
- anthropic
- anthropic:default static
- claude-cli
- anthropic:claude-cli expiring expires in 5h
- openai-codex usage: 5h 91% left β±2h 16m Β· Week 88% left β±5d 8h
- openai-codex:default ok expires in 6d
- openai-codex:majorkarpati@gmail.com (majorkarpati@gmail.com) ok expires in 10d
π¦ OpenClaw 2026.4.25 (aa36ee6) β I don't just autocompleteβI auto-commit (emotionally), then ask you to review (logically).
β
β
Service: systemd (enabled)
File logs: /tmp/openclaw/openclaw-2026-04-29.log
Command: /usr/bin/node /home/major/.npm-global/lib/node_modules/openclaw/dist/index.js gateway --port 18789
Service file: ~/.config/systemd/user/openclaw-gateway.service
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 3057569, state active, sub running, last exit 0, reason 0)
Connectivity probe: ok
Capability: admin-capable
Listening: [::1]:18789
Troubles: run openclaw status
Troubleshooting: https://docs.openclaw.ai/troubleshooting
OK das war der ouput
Die Config sieht grundsΓ€tzlich ok aus: openai-codex/gpt-5.5 ist als Default gesetzt, in den configured models drin, und openai-codex OAuth ist gΓΌltig. Also: ja, OpenClaw benutzt bei dir gerade Codex/GPTβ5.5 als Default.
Das Problem ist eher: Gateway/Session hΓ€ngt. In deinem Output stand erst Gateway health: timeout / chat.history timeout, spΓ€ter aber Connectivity probe: ok. Ich wΓΌrde jetzt so vorgehen:
openclaw gateway restart
openclaw gateway status
Dann in einer frischen Session kurz testen:
openclaw infer model run --model openai-codex/gpt-5.5 --prompt "Sag nur: ok" --json
Wenn das klappt, ist Codex 5.5 selbst okay und nur deine alte Main-Session hΓ€ngt. Dann im UI/Chat die hΓ€ngende Session stoppen (/stop oder Stop-Button) und frisch weitermachen.
AuΓerdem: dein Status zeigt Update auf 2026.4.26. Ich wΓΌrde updaten, nicht openclaw update verwenden:
pnpm add -g openclaw@latest
openclaw gateway install --force
openclaw gateway restart
Falls danach noch keine Reaktion kommt, poste bitte den relevanten Teil aus:
openclaw logs --follow
direkt nachdem du eine Nachricht an die Main-Session schickst.
Ich hab mir die logs angeschaut und das sieht nicht gut aus.
1/10
Log file: /tmp/openclaw/openclaw-2026-04-29.log
20:56:11+00:00 info gateway {"subsystem":"gateway"} restart mode: full process restart (supervisor restart)
20:56:21+00:00 info gateway {"subsystem":"gateway"} loading configurationβ¦
20:56:21+00:00 info gateway {"subsystem":"gateway"} resolving authenticationβ¦
20:56:21+00:00 info gateway {"subsystem":"gateway"} starting...
20:56:30+00:00 info gateway {"subsystem":"gateway"} starting HTTP server...
20:56:30+00:00 info gateway/canvas {"subsystem":"gateway/canvas"} canvas host mounted at http://127.0.0.1:18789/__openclaw__/canvas/ (root /home/major/.openclaw/canvas)
20:56:30+00:00 info gateway/health-monitor {"subsystem":"gateway/health-monitor"} started (interval: 300s, startup-grace: 60s, channel-connect-grace: 120s)
20:56:31+00:00 info gateway {"subsystem":"gateway"} agent model: anthropic/claude-opus-4-6
20:56:31+00:00 info gateway {"subsystem":"gateway"} ready (3 plugins: anthropic, browser, telegram; 9.7s)
20:56:31+00:00 info gateway {"subsystem":"gateway"} log file: /tmp/openclaw/openclaw-2026-04-29.log
20:56:31+00:00 info gateway {"subsystem":"gateway"} starting channels and sidecars...
20:56:31+00:00 info gateway/hooks {"subsystem":"gateway/hooks"} loaded 4 internal hook handlers
20:57:45+00:00 info browser/server {"subsystem":"browser/server"} Browser control listening on http://127.0.0.1:18791/ (auth=token)
20:58:09+00:00 warn gateway/ws {"subsystem":"gateway/ws"} handshake timeout conn=43c48ea7-2b47-4117-9e22-d6e896125bb6 peer=127.0.0.1:42714->127.0.0.1:18789 remote=127.0.0.1
20:58:09+00:00 info gateway/channels/telegram {"subsystem":"gateway/channels/telegram"} [bewerbungen] starting provider
20:58:09+00:00 info gateway/channels/telegram {"subsystem":"gateway/channels/telegram"} [default] starting provider
Der erste Block ist noch nicht βkaputtβ genug fΓΌr eine Diagnose β Gateway startet sauber, aber auffΓ€llig sind zwei Dinge:
agent model: anthropic/claude-opus-4-6beim Start, obwohl deinmodels statusspΓ€teropenai-codex/gpt-5.5zeigt.gateway/ws handshake timeoutpasst zu deinemchat.history timeout.
Schick ruhig die restlichen Teile; besonders spannend sind Zeilen mit error, timeout, session, chat.history, openai-codex, gpt-5.5 oder incomplete turn.
2/10
20:58:09+00:00 info gateway/channels/telegram {"subsystem":"gateway/channels/telegram"} [trading] starting provider
20:58:09+00:00 info gateway/channels/telegram {"subsystem":"gateway/channels/telegram"} [pokemon] starting provider
20:58:09+00:00 info gateway/channels/telegram {"subsystem":"gateway/channels/telegram"} [voidbloom] starting provider
20:58:09+00:00 info gateway/heartbeat {"subsystem":"gateway/heartbeat"} {"intervalMs":2700000} heartbeat: started
20:58:09+00:00 info gateway {"subsystem":"gateway"} signal SIGTERM received
20:58:09+00:00 info gateway {"subsystem":"gateway"} received SIGTERM; restarting
20:58:16+00:00 error gateway connect failed: Error: gateway closed (1006):
20:58:20+00:00 info gateway {"subsystem":"gateway"} loading configurationβ¦
20:58:20+00:00 info gateway {"subsystem":"gateway"} resolving authenticationβ¦
20:58:20+00:00 info gateway {"subsystem":"gateway"} starting...
20:58:28+00:00 info gateway {"subsystem":"gateway"} starting HTTP server...
20:58:28+00:00 info gateway/canvas {"subsystem":"gateway/canvas"} canvas host mounted at http://127.0.0.1:18789/__openclaw__/canvas/ (root /home/major/.openclaw/canvas)
20:58:28+00:00 info gateway/health-monitor {"subsystem":"gateway/health-monitor"} started (interval: 300s, startup-grace: 60s, channel-connect-grace: 120s)
3/10
20:58:28+00:00 info gateway {"subsystem":"gateway"} agent model: anthropic/claude-opus-4-6
20:58:28+00:00 info gateway {"subsystem":"gateway"} ready (3 plugins: anthropic, browser, telegram; 8.1s)
20:58:28+00:00 info gateway {"subsystem":"gateway"} log file: /tmp/openclaw/openclaw-2026-04-29.log
20:58:28+00:00 info gateway {"subsystem":"gateway"} starting channels and sidecars...
20:58:28+00:00 info gateway/hooks {"subsystem":"gateway/hooks"} loaded 4 internal hook handlers
20:59:39+00:00 info browser/server {"subsystem":"browser/server"} Browser control listening on http://127.0.0.1:18791/ (auth=token)
21:00:02+00:00 warn gateway/ws {"subsystem":"gateway/ws"} handshake timeout conn=ef2de2fb-c56e-40f7-99a4-f6c153ebd022 peer=127.0.0.1:42032->127.0.0.1:18789 remote=127.0.0.1
21:00:02+00:00 info gateway/channels/telegram {"subsystem":"gateway/channels/telegram"} [bewerbungen] starting provider
21:00:03+00:00 info gateway/channels/telegram {"subsystem":"gateway/channels/telegram"} [default] starting provider
21:00:03+00:00 info gateway/channels/telegram {"subsystem":"gateway/channels/telegram"} [trading] starting provider
21:00:03+00:00 info gateway/channels/telegram {"subsystem":"gateway/channels/telegram"} [voidbloom] starting provider
21:00:03+00:00 info gateway/channels/telegram {"subsystem":"gateway/channels/telegram"} [pokemon] starting provider
21:00:03+00:00 info gateway/heartbeat {"subsystem":"gateway/heartbeat"} {"intervalMs":2700000} heartbeat: started
4/10
1:00:03+00:00 warn gateway/ws {"subsystem":"gateway/ws"} {"cause":"handshake-timeout","handshake":"failed","durationMs":23619,"host":"127.0.0.1:18789","origin":"http://127.0.0.1:18789","userAgent":"Mozilla/5.0 (X11; Linux x86_64; rv:147.0) Gecko/20100101 Firefox/147.0","remoteAddr":"127.0.0.1","remotePort":42032,"localAddr":"127.0.0.1","localPort":18789,"endpoint":"127.0.0.1:42032->127.0.0.1:18789","handshakeMs":23380} closed before connect conn=ef2de2fb-c56e-40f7-99a4-f6c153ebd022 peer=127.0.0.1:42032->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 (X11; Linux x86_64; rv:147.0) Gecko/20100101 Firefox/147.0 code=1006 reason=n/a
21:00:03+00:00 warn memory {"subsystem":"memory"} qmd memory unavailable; falling back to builtin: Cannot find package 'chokidar' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.25-de49e110caae/dist/qmd-manager-FuXCtSYP.js
Did you mean to import "chokidar/index.js"?
21:00:03+00:00 warn gateway {"subsystem":"gateway"} qmd memory startup initialization failed for agent "main": Cannot find package 'chokidar' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.25-de49e110caae/dist/manager-C6O7HXGc.js
Did you mean to import "chokidar/index.js"?
21:00:03+00:00 warn memory {"subsystem":"memory"} qmd memory unavailable; falling back to builtin: Cannot find package 'chokidar' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.25-de49e110caae/dist/qmd-manager-FuXCtSYP.js
Did you mean to import "chokidar/index.js"?
21:00:03+00:00 warn gateway {"subsystem":"gateway"} qmd memory startup initialization failed for agent "bewerbungen": Cannot find package 'chokidar' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.25-de49e110caae/dist/manager-C6O7HXGc.js
Did you mean to import "chokidar/index.js"?
5/10
21:00:03+00:00 info gateway/channels/telegram {"subsystem":"gateway/channels/telegram"} Telegram menu text exceeded the conservative 5700-character payload budget; shortening descriptions to keep 60 commands visible.
21:00:04+00:00 info gateway/channels/telegram {"subsystem":"gateway/channels/telegram"} Telegram menu text exceeded the conservative 5700-character payload budget; shortening descriptions to keep 60 commands visible.
21:00:04+00:00 info gateway/channels/telegram {"subsystem":"gateway/channels/telegram"} Telegram menu text exceeded the conservative 5700-character payload budget; shortening descriptions to keep 60 commands visible.
21:00:04+00:00 info gateway/channels/telegram {"subsystem":"gateway/channels/telegram"} Telegram menu text exceeded the conservative 5700-character payload budget; shortening descriptions to keep 60 commands visible.
21:00:04+00:00 warn gateway/ws {"subsystem":"gateway/ws"} {"handshake":"pending","durationMs":1337,"host":"127.0.0.1:18789","origin":"http://127.0.0.1:18789","userAgent":"Mozilla/5.0 (X11; Linux x86_64; rv:147.0) Gecko/20100101 Firefox/147.0","remoteAddr":"127.0.0.1","remotePort":42266,"localAddr":"127.0.0.1","localPort":18789,"endpoint":"127.0.0.1:42266->127.0.0.1:18789"} closed before connect conn=7c92018e-0dfb-4f47-be44-bfdfbf35191a peer=127.0.0.1:42266->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 (X11; Linux x86_64; rv:147.0) Gecko/20100101 Firefox/147.0 code=1006 reason=n/a
21:00:04+00:00 info gateway/channels/telegram {"subsystem":"gateway/channels/telegram"} Telegram menu text exceeded the conservative 5700-character payload budget; shortening descriptions to keep 60 commands visible.
21:00:04+00:00 warn memory {"subsystem":"memory"} qmd memory unavailable; falling back to builtin: Cannot find package 'chokidar' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.25-de49e110caae/dist/qmd-manager-FuXCtSYP.js
6/10
Did you mean to import "chokidar/index.js"?
21:00:04+00:00 warn gateway {"subsystem":"gateway"} qmd memory startup initialization failed for agent "trading": Cannot find package 'chokidar' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.25-de49e110caae/dist/manager-C6O7HXGc.js
Did you mean to import "chokidar/index.js"?
21:00:04+00:00 info cron {"module":"cron","storePath":"/home/major/.openclaw/cron/jobs.json"} {"enabled":true,"jobs":1,"nextWakeAtMs":null} cron: started
21:00:04+00:00 warn memory {"subsystem":"memory"} qmd memory unavailable; falling back to builtin: Cannot find package 'chokidar' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.25-de49e110caae/dist/qmd-manager-FuXCtSYP.js
Did you mean to import "chokidar/index.js"?
21:00:04+00:00 warn gateway {"subsystem":"gateway"} qmd memory startup initialization failed for agent "voidbloom": Cannot find package 'chokidar' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.25-de49e110caae/dist/manager-C6O7HXGc.js
Did you mean to import "chokidar/index.js"?
21:00:04+00:00 warn memory {"subsystem":"memory"} qmd memory unavailable; falling back to builtin: Cannot find package 'chokidar' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.25-de49e110caae/dist/qmd-manager-FuXCtSYP.js
Did you mean to import "chokidar/index.js"?
21:00:04+00:00 warn gateway {"subsystem":"gateway"} qmd memory startup initialization failed for agent "pokemon": Cannot find package 'chokidar' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.25-de49e110caae/dist/manager-C6O7HXGc.js
Did you mean to import "chokidar/index.js"?
7/10
21:00:05+00:00 info gateway/ws {"subsystem":"gateway/ws"} webchat connected conn=2540b36f-f6c8-490e-86d4-3fa7894c9bad remote=127.0.0.1 client=openclaw-control-ui webchat v2026.4.14
21:00:44+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β chat.history 39557ms conn=ec13cb92β¦73e1 id=6d570a98β¦9d6e
21:00:44+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β sessions.list 84ms conn=2540b36fβ¦9bad id=c5fdbc26β¦ca4b
21:00:45+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β commands.list 184ms conn=2540b36fβ¦9bad id=0f9bfa1cβ¦24e3
21:00:45+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β chat.history 342ms conn=2540b36fβ¦9bad id=a1d13b64β¦e311
21:00:45+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β models.list 199ms conn=2540b36fβ¦9bad id=e280bea8β¦bd8b
21:00:45+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β node.list 602ms conn=2540b36fβ¦9bad id=e25b8fe4β¦7359
21:00:45+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β device.pair.list 607ms conn=2540b36fβ¦9bad id=3a97a12dβ¦5cb2
21:00:45+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β health 734ms conn=2540b36fβ¦9bad id=226a6442β¦7567
21:00:52+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β node.list 527ms conn=2540b36fβ¦9bad id=ea27d60dβ¦5e3a
21:00:52+00:00 info agent/cli-backend {"subsystem":"agent/cli-backend"} cli session reset: provider=claude-cli reason=auth-profile
21:00:53+00:00 info agent/cli-backend {"subsystem":"agent/cli-backend"} cli exec: provider=claude-cli model=opus promptChars=205
21:00:53+00:00 info agent/cli-backend {"subsystem":"agent/cli-backend"} claude live session start: provider=claude-cli model=claude-opus-4-6 activeSessions=1
21:02:19+00:00 warn agent/cli-backend {"subsystem":"agent/cli-backend"} claude live session turn failed: provider=claude-cli model=claude-opus-4-6 durationMs=86324 error=FailoverError
21:02:20+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β node.list 367ms conn=2540b36fβ¦9bad id=0d99f2a6β¦103c
21:02:20+00:00 info agent/cli-backend {"subsystem":"agent/cli-backend"} cli exec: provider=claude-cli model=opus promptChars=384
21:02:20+00:00 info agent/cli-backend {"subsystem":"agent/cli-backend"} claude live session start: provider=claude-cli model=claude-opus-4-6 activeSessions=2
21:03:08+00:00 warn diagnostic {"subsystem":"diagnostic"} stuck session: sessionId=unknown sessionKey=*** state=processing age=143s queueDepth=1
21:03:08+00:00 warn diagnostic {"subsystem":"diagnostic"} stuck session: sessionId=unknown sessionKey=*** state=processing age=143s queueDepth=1
21:03:08+00:00 info typing TTL reached (2m); stopping typing indicator
21:03:08+00:00 warn agent/cli-backend {"subsystem":"agent/cli-backend"} claude live session turn failed: provider=claude-cli model=claude-opus-4-6 durationMs=47879 error=FailoverError
21:03:23+00:00 warn gateway/ws {"subsystem":"gateway/ws"} handshake timeout conn=b6db5f0b-d024-4c00-973f-1f1e19971205 peer=127.0.0.1:48060->127.0.0.1:18789 remote=127.0.0.1
21:03:23+00:00 warn gateway/ws {"subsystem":"gateway/ws"} {"cause":"handshake-timeout","handshake":"failed","durationMs":15533,"host":"127.0.0.1:18789","remoteAddr":"127.0.0.1","remotePort":48060,"localAddr":"127.0.0.1","localPort":18789,"endpoint":"127.0.0.1:48060->127.0.0.1:18789","handshakeMs":15523} closed before connect conn=b6db5f0b-d024-4c00-973f-1f1e19971205 peer=127.0.0.1:48060->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=1000 reason=n/a
21:03:23+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β node.list 15608ms conn=2540b36fβ¦9bad id=8bb9e23cβ¦796b
21:03:23+00:00 error gateway connect failed: Error: gateway closed (1000):
21:04:01+00:00 warn diagnostic {"subsystem":"diagnostic"} stuck session: sessionId=unknown sessionKey=*** state=processing age=197s queueDepth=1
21:04:01+00:00 warn diagnostic {"subsystem":"diagnostic"} stuck session: sessionId=unknown sessionKey=*** state=processing age=197s queueDepth=1
21:04:01+00:00 info gateway/ws {"subsystem":"gateway/ws"} webchat disconnected code=1001 reason=n/a conn=2540b36f-f6c8-490e-86d4-3fa7894c9bad
21:04:02+00:00 warn model-fallback/decision {"subsystem":"model-fallback/decision"} {"event":"model_fallback_decision","tags":["error_handling","model_fallback","candidate_failed"],"runId":"070ad35f-70d6-439d-9909-f5f7800c0bc4","decision":"candidate_failed","requestedProvider":"anthropic","requestedModel":"claude-opus-4-6","candidateProvider":"anthropic","candidateModel":"claude-opus-4-6","attempt":1,"total":3,"reason":"unknown","errorPreview":"Your organization does not have access to Claude. Please login again or contact your administrator.","errorHash":"sha256:e9763ff27995","nextCandidateProvider":"openai-codex","nextCandidateModel":"gpt-5.4","isPrimary":true,"requestedModelMatched":true,"fallbackConfigured":true} model fallback decision
21:04:02+00:00 warn gateway/ws {"subsystem":"gateway/ws"} {"handshake":"pending","durationMs":96,"host":"127.0.0.1:18789","origin":"http://127.0.0.1:18789","userAgent":"Mozilla/5.0 (X11; Linux x86_64; rv:147.0) Gecko/20100101 Firefox/147.0","remoteAddr":"127.0.0.1","remotePort":42862,"localAddr":"127.0.0.1","localPort":18789,"endpoint":"127.0.0.1:42862->127.0.0.1:18789"} closed before connect conn=98def304-3fb3-43e6-a9ab-426033a2aad6 peer=127.0.0.1:42862->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 (X11; Linux x86_64; rv:147.0) Gecko/20100101 Firefox/147.0 code=1006 reason=n/a
21:04:02+00:00 warn agent/embedded {"subsystem":"agent/embedded"} session file repaired: rewrote 2 assistant message(s) (42c66412-741a-4e69-98b2-4ea4723c9e66.jsonl)
21:04:34+00:00 warn gateway/ws {"subsystem":"gateway/ws"} handshake timeout conn=995d0654-d387-44be-98e3-edc3567e8f3f peer=127.0.0.1:35464->127.0.0.1:18789 remote=127.0.0.1
21:04:34+00:00 warn diagnostic {"subsystem":"diagnostic"} stuck session: sessionId=unknown sessionKey=*** state=processing age=229s queueDepth=1
21:04:34+00:00 warn gateway/ws {"subsystem":"gateway/ws"} {"cause":"handshake-timeout","handshake":"failed","durationMs":30973,"lastFrameType":"req","lastFrameMethod":"connect","lastFrameId":"c7370bd8-d8c3-4b51-a6c9-b1be84193141","host":"127.0.0.1:18789","origin":"http://127.0.0.1:18789","userAgent":"Mozilla/5.0 (X11; Linux x86_64; rv:147.0) Gecko/20100101 Firefox/147.0","remoteAddr":"127.0.0.1","remotePort":35464,"localAddr":"127.0.0.1","localPort":18789,"endpoint":"127.0.0.1:35464->127.0.0.1:18789","handshakeMs":30938} closed before connect conn=995d0654-d387-44be-98e3-edc3567e8f3f peer=127.0.0.1:35464->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 (X11; Linux x86_64; rv:147.0) Gecko/20100101 Firefox/147.0 code=1000 reason=n/a
21:04:34+00:00 info gateway/ws {"subsystem":"gateway/ws"} webchat connected conn=995d0654-d387-44be-98e3-edc3567e8f3f remote=127.0.0.1 client=openclaw-control-ui webchat vcontrol-ui
21:04:36+00:00 info gateway/ws {"subsystem":"gateway/ws"} webchat connected conn=692f72e5-4b00-4b18-8fd7-26ab40e8e54e remote=127.0.0.1 client=openclaw-control-ui webchat vcontrol-ui
21:04:36+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β commands.list 265ms conn=692f72e5β¦e54e id=0ee24864β¦079a
21:04:36+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β chat.history 387ms conn=692f72e5β¦e54e id=877c9a64β¦489e
21:04:36+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β models.list 297ms conn=692f72e5β¦e54e id=c2fa24ecβ¦7e32
21:04:36+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β node.list 438ms conn=692f72e5β¦e54e id=7169967aβ¦0ef1
21:04:36+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β device.pair.list 447ms conn=692f72e5β¦e54e id=31ff0f42β¦3861
21:04:41+00:00 error [tools] read failed: ENOENT: no such file or directory, access '/home/major/.openclaw/workspace/memory/2026-04-29.md' raw_params={"path":"/home/major/.openclaw/workspace/memory/2026-04-29.md"}
21:04:41+00:00 error [tools] read failed: ENOENT: no such file or directory, access '/home/major/.openclaw/workspace/memory/2026-04-29.md' raw_params={"path":"/home/major/.openclaw/workspace/memory/2026-04-29.md"}
21:04:41+00:00 error [tools] read failed: ENOENT: no such file or directory, access '/home/major/.openclaw/workspace/memory/2026-04-28.md' raw_params={"path":"/home/major/.openclaw/workspace/memory/2026-04-28.md"}
21:04:41+00:00 error [tools] read failed: ENOENT: no such file or directory, access '/home/major/.openclaw/workspace/memory/2026-04-28.md' raw_params={"path":"/home/major/.openclaw/workspace/memory/2026-04-28.md"}
21:04:46+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β sessions.list 65ms conn=692f72e5β¦e54e id=0304e2c0β¦50b1
21:04:47+00:00 warn model-fallback/decision {"subsystem":"model-fallback/decision"} {"event":"model_fallback_decision","tags":["error_handling","model_fallback","candidate_succeeded"],"runId":"34678ad8-7c88-42e0-afbb-5557a7dfc647","decision":"candidate_succeeded","requestedProvider":"anthropic","requestedModel":"claude-opus-4-6","candidateProvider":"openai-codex","candidateModel":"gpt-5.4","attempt":2,"total":3,"isPrimary":false,"requestedModelMatched":false,"fallbackConfigured":true,"previousAttempts":[{"provider":"anthropic","model":"claude-opus-4-6","reason":"unknown","errorPreview":"Your organization does not have access to Claude. Please login again or contact your administrator.","errorHash":"sha256:e9763ff27995"}]} model fallback decision
21:04:47+00:00 info gateway {"subsystem":"gateway"} signal SIGTERM received
21:04:47+00:00 info gateway {"subsystem":"gateway"} received SIGTERM; restarting
21:04:47+00:00 info gateway {"subsystem":"gateway"} received SIGTERM; restarting
21:04:47+00:00 info gateway {"subsystem":"gateway"} draining 2 active task(s) and 1 active embedded run(s) before restart without a timeout
21:05:04+00:00 warn Config overwrite: /home/major/.openclaw/openclaw.json (sha256 2b916587e61817dd9e995415dba9d10bd209d5aa845ec91eb808c9e8926b038e -> b94053d1a264643eed00e2fa09d8cf2df0298f9e6a95990ca6c803874ebc4fb8, backup=/home/major/.openclaw/openclaw.json.bak)
21:05:05+00:00 info Updated agents.defaults.model.primary. Restart the gateway to apply.
21:05:05+00:00 info gateway/reload {"subsystem":"gateway/reload"} config change detected; evaluating reload (meta.lastTouchedAt, agents.defaults.model.primary)
21:05:05+00:00 info gateway/reload {"subsystem":"gateway/reload"} config hot reload applied (agents.defaults.model.primary)
21:05:13+00:00 info gateway/ws {"subsystem":"gateway/ws"} β res β sessions.list 73ms conn=692f72e5β¦e54e id=7a2076b4β¦d537
21:05:32+00:00 warn gateway {"subsystem":"gateway"} still draining 2 active task(s) and 1 active embedded run(s) before restart
21:05:52+00:00 info gateway {"subsystem":"gateway"} loading configurationβ¦
21:05:52+00:00 info gateway {"subsystem":"gateway"} resolving authenticationβ¦
21:05:52+00:00 info gateway {"subsystem":"gateway"} starting...
Das reicht schon β da sind die wichtigen Hinweise drin.
Hauptproblem: Deine Session lΓ€uft gerade nicht sauber auf Codex 5.5, sondern versucht noch claude-cli / claude-opus-4-6:
cli exec: provider=claude-cli model=opus
...
Your organization does not have access to Claude
...
nextCandidateProvider: openai-codex
nextCandidateModel: gpt-5.4
Also: Claude failt, dann fΓ€llt OpenClaw auf Codex 5.4 zurΓΌck. Deine hΓ€ngende Main-Session hat vermutlich noch ein altes Session-Override/Fallback-State.
Mach jetzt bitte:
pnpm add -g openclaw@latest
openclaw gateway install --force
openclaw gateway restart
Dann im Browser: Control UI hart neu laden / Site Data fΓΌr 127.0.0.1:18789 lΓΆschen. In den Logs steht nΓ€mlich webchat v2026.4.14 gegen Gateway 2026.4.25 β das ist verdΓ€chtig alt.
Danach in einer neuen Session testen:
/model openai-codex/gpt-5.5
oder direkt:
openclaw infer model run --model openai-codex/gpt-5.5 --prompt "Sag nur: ok" --json
Die chokidar/plugin-runtime-deps Warnungen sind auch unschΓΆn, aber vermutlich nicht der direkte Grund fΓΌr die fehlende Antwort β Update + Gateway reinstall sollte das als erstes bereinigen.
Docs dazu: https://docs.openclaw.ai/concepts/models und https://docs.openclaw.ai/gateway/troubleshooting
Ja, der letzte Block bestΓ€tigt es nochmal: die UI-Verbindung erholt sich teilweise (chat.history 387ms), aber die eigentliche Session bleibt stuck:
stuck session ... state=processing age=229s queueDepth=1
Ich wΓΌrde nicht weiter in dieser alten Main-Session debuggen. Recovery:
-
Gateway + Version sauber aktualisieren:
pnpm add -g openclaw@latest openclaw gateway install --force openclaw gateway restart -
Browser-Control-UI hart refreshen / Site Data fΓΌr
127.0.0.1:18789lΓΆschen.
-
Neue Session starten und dort explizit setzen:
/model openai-codex/gpt-5.5 -
Falls die alte Session weiter hΓ€ngt: im UI Stop drΓΌcken oder
/stop. Wenn sie danach immer noch alsprocessinghΓ€ngt, ist das wahrscheinlich ein Session-State-Bug β dann bitte als GitHub Issue mit den Logstellen einreichen: https://github.com/openclaw/openclaw/issues
Nebenbefund: dein Claude auth/profile ist kaputt oder hat keinen Org-Zugriff (Your organization does not have access to Claude). Wenn du wirklich nur Codex nutzen willst, wΓΌrde ich Claude als Fallback vorerst rausnehmen, damit er nicht erst 1β2 Minuten in Claude-Fehler lΓ€uft.
Yep, this block nails it:
ENOENT memory/2026-04-29.md= mostly harmless. Agent tried to read daily memory files that donβt exist.- Important part:
So that run was not using Codex 5.5. It started on Claude, failed, then fell back to Codex 5.4.requestedModel: claude-opus-4-6 error: Your organization does not have access to Claude candidate_succeeded: openai-codex/gpt-5.4
Since your later models status already shows default openai-codex/gpt-5.5, Iβd treat the old main session as poisoned/stale.
Do this now:
openclaw gateway restart
Then start a fresh session:
/new openai-codex/gpt-5.5
or inside a new session:
/model openai-codex/gpt-5.5
If Claude is still in fallbacks and you donβt need it, remove/disable that fallback later so runs donβt waste 1β2 minutes failing Claude first. The stuck old session + repaired session file strongly suggests not debugging that one further.
Das ist genau der Moment, wo deine Γnderung auf agents.defaults.model.primary passiert ist. Aber: der Gateway musste noch 2 aktive Tasks + 1 embedded run drainen, also liefen alte Runs weiter und konnten noch mit dem alten Claude/5.4-Fallback-State arbeiten.
Wichtig: nach 21:05:52 ... starting sollten wir sehen, welches Modell der Gateway beim Neustart lΓ€dt. Such bitte ab da nach:
agent model:
Wenn dort openai-codex/gpt-5.5 steht, ist die Default-Config jetzt aktiv.
Wenn dort weiter anthropic/claude-opus-4-6 steht, dann gibt es sehr wahrscheinlich ein Agent-spezifisches Override oder Session-Override, das agents.defaults.model.primary ΓΌberstimmt. Dann brauchen wir:
openclaw config get agents.list
openclaw config get agents.defaults.model
Und nochmal: die alte hΓ€ngende Session wΓΌrde ich nicht weiterverwenden β neue Session mit /new openai-codex/gpt-5.5 ist der sauberste Test.
Ok danke ich hatte es jetzt kurz am laufen jetzt, bekomme ich aber im tui gateway connect failed: Error: gateway request timeout for connect
Das ist wieder Gateway-seitig, nicht Codex selbst.
Mach bitte einmal den βhart aber sauberβ-Restart, weil dein Log vorhin draining ... without a timeout gezeigt hat:
openclaw gateway stop
sleep 3
openclaw gateway start
openclaw gateway status
openclaw status --all
Wenn gateway stop hΓ€ngt oder danach weiter request timeout for connect kommt:
systemctl --user restart openclaw-gateway
sleep 5
openclaw gateway status
Dann nochmal TUI starten.
Wenn es weiter passiert, bitte diese zwei Sachen posten:
openclaw --version
openclaw gateway status --deep
tail -n 120 /tmp/openclaw/openclaw-2026-04-29.log
Wichtig: Falls der TUI nicht auf derselben Maschine lΓ€uft wie der Gateway, ist 127.0.0.1:18789 falsch β das wΓ€re dann der lokale Rechner des TUI, nicht der Gateway-Host.
ajor@localhost-live:~$ openclaw gateway stop
sleep 3
openclaw gateway start
openclaw gateway status
openclaw status --all
π¦ OpenClaw 2026.4.26 (be8c246) β I've read more man pages than any human shouldβso you don't have to.
Stopped systemd service: openclaw-gateway.service
π¦ OpenClaw 2026.4.26 (be8c246) β More integrations than your therapist's intake form.
Restarted systemd service: openclaw-gateway.service
π¦ OpenClaw 2026.4.26 (be8c246) β I've seen your commit messages. We'll work on that together.
β
β
Service: systemd (enabled)
File logs: /tmp/openclaw/openclaw-2026-04-30.log
Command: /usr/bin/node /home/major/.npm-global/lib/node_modules/openclaw/dist/index.js gateway --port 18789
Service file: ~/.config/systemd/user/openclaw-gateway.service
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 3062759, state active, sub running, last exit 0, reason 0)
Warm-up: launch agents can take a few seconds. Try again shortly.
Connectivity probe: failed
Probe target: ws://127.0.0.1:18789
connect ECONNREFUSED 127.0.0.1:18789
Capability: unknown
Gateway port 18789 is not listening (service appears running).
Logs: journalctl --user -u openclaw-gateway.service -n 200 --no-pager
Restart log: ~/.openclaw/logs/gateway-restart.log
Troubles: run openclaw status
Troubleshooting: https://docs.openclaw.ai/troubleshooting
Das sieht erstmal nach zu frΓΌh nach dem Start geprΓΌft aus β Warm-up + ECONNREFUSED heiΓt: systemd-Prozess lΓ€uft, aber Gateway hat den HTTP/WS-Port noch nicht geΓΆffnet.
Bitte einmal:
sleep 20
openclaw gateway status
Wenn dann immer noch βport 18789 is not listeningβ kommt, brauchen wir die Startfehler:
journalctl --user -u openclaw-gateway.service -n 200 --no-pager
tail -n 200 /tmp/openclaw/openclaw-2026-04-30.log
cat ~/.openclaw/logs/gateway-restart.log
Wichtig: Du bist jetzt auf 2026.4.26, gut. Wenn die Logs wieder Cannot find package 'chokidar' oder Plugin-Runtime-Deps Fehler zeigen, wΓ€re der nΓ€chste Schritt vermutlich:
openclaw doctor
openclaw gateway restart
Aber erst schauen wir, ob der Port nach 20β30s hochkommt.
π¦ OpenClaw 2026.4.26 (be8c246) β If something's on fire, I can't extinguish itβbut I can write a beautiful postmortem.
β
OpenClaw status --all
Overview
ββββββββββββββββββββββ¬ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β Item β Value β
ββββββββββββββββββββββΌββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β Version β 2026.4.26 β
β OS β linux 6.18.8-200.fc43.x86_64 (x64) β
β Node β 22.22.0 β
β Config β ~/.openclaw/openclaw.json β
β Dashboard β http://127.0.0.1:18789/ β
β Tailscale exposure β off β
β Channel β stable (config) β
β Update β pnpm Β· npm update 2026.4.27
[30.04.26 01:36] Tobias: β
β Gateway β local Β· ws://127.0.0.1:18789 (local loopback) Β· unreachable (connect ECONNREFUSED 127.0.0.1:18789) β
β Security β Run: openclaw security audit --deep β
β Gateway self β unknown β
β Gateway service β systemd installed Β· enabled Β· running (pid 3062759, state active) β
β Node service β systemd not installed β
β Agents β 5 total Β· 2 bootstrapping Β· 1 active Β· 13 sessions β
β Secrets β none β
ββββββββββββββββββββββ΄ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
Channels
ββββββββββββ¬ββββββββββ¬βββββββββ¬βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β Channel β Enabled β State β Detail β
ββββββββββββΌββββββββββΌβββββββββΌβββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
ββββββββββββ΄ββββββββββ΄βββββββββ΄βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
Agents
βββββββββββββββββββββββββββββ¬βββββββββββββββββ¬βββββββββββ¬βββββββββββ¬ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β Agent β Bootstrap file β Sessions β Active β Store β
βββββββββββββββββββββββββββββΌβββββββββββββββββΌβββββββββββΌβββββββββββΌββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β main β ABSENT β 5 β 6m ago β ~/.openclaw/agents/main/sessions/sessions.json β
β bewerbungen (bewerbungen) β PRESENT β 2 β 14d ago β ~/.openclaw/agents/bewerbungen/sessions/sessions.json β
β pokemon (pokemon) β ABSENT β 2 β 2h ago β ~/.openclaw/agents/pokemon/sessions/sessions.json β
β trading (trading) β ABSENT β 2 β 2h ago β ~/.openclaw/agents/trading/sessions/sessions.jsonβ
β voidbloom (voidbloom) β PRESENT β 2 β 22d ago β ~/.openclaw/agents/voidbloom/sessions/sessions.json β
βββββββββββββββββββββββββββββ΄βββββββββββββββββ΄βββββββββββ΄βββββββββββ΄ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
Diagnosis (read-only)
Okay, dann ist es nicht nur Warm-up. systemd sagt βrunningβ, aber der Gateway ΓΆffnet den Port nicht β Start hΓ€ngt oder crasht vor HTTP bind.
Bitte jetzt genau diese Logs:
journalctl --user -u openclaw-gateway.service -n 200 --no-pager
tail -n 200 /tmp/openclaw/openclaw-2026-04-30.log
cat ~/.openclaw/logs/gateway-restart.log
Und zusΓ€tzlich einmal:
openclaw doctor
Wenn du schnell testen willst, ob es wieder der Plugin-Runtime-Deps/chokidar Kram ist, achte in den Logs auf sowas:
Cannot find package 'chokidar'
plugin-runtime-deps
Wenn das auftaucht, sag Bescheid β dann wΓΌrde ich als nΓ€chsten Schritt die Runtime-Deps neu aufbauen lassen, statt weiter an Codex rumzuschrauben.
Gateway connection details:
Gateway target: ws://127.0.0.1:18789
Source: local loopback
Config: /home/major/.openclaw/openclaw.json
Bind: loopback
β Config: /home/major/.openclaw/openclaw.json
β Secret diagnostics (0)
β Restart sentinel: none
β Port 18789
! Tailscale exposure: off Β· daemon unknown
β Skills: 17 eligible Β· 0 missing Β· /home/major/.openclaw/workspace
β Plugin compatibility (none)
! 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
Danke β der Status bestΓ€tigt nur nochmal: Gateway ist als Service da, aber nicht erreichbar. Wir brauchen jetzt wirklich die Journal-/Gateway-Logs, weil da der eigentliche Start-HΓ€nger drinsteht.
Bitte posten:
journalctl --user -u openclaw-gateway.service -n 200 --no-pager
tail -n 200 /tmp/openclaw/openclaw-2026-04-30.log
cat ~/.openclaw/logs/gateway-restart.log
Wennβs viel ist, wieder in Teilen schicken. Die interessantesten Zeilen sind direkt nach loading configurationβ¦ / resolving authenticationβ¦ / startingβ¦.
Yep β ECONNREFUSED means nichts lauscht auf 18789, obwohl systemd noch einen Prozess meldet. Das ist Startphase/HΓ€nger vor dem HTTP server, nicht ein normaler Auth-/Codex-Fehler.
NΓ€chster Schritt bleibt:
journalctl --user -u openclaw-gateway.service -n 200 --no-pager
tail -n 200 /tmp/openclaw/openclaw-2026-04-30.log
cat ~/.openclaw/logs/gateway-restart.log
Ohne diese Logs wΓΌrden wir nur raten.
Ja also, es scheint jetzt wieder zu laufen, allerdings bricht die Verbindung manchmal ab und dann bekomme ich im terminal folgende Meldung send failed: Error: also es ist komisch.
Ja, das passt zu einem instabilen Gateway/WebSocket, nicht mehr direkt zu Codex.
Wenn es nach Gateway-Restart kurz geht und dann send failed: Error: kommt, mach bitte als nΓ€chstes:
pnpm add -g openclaw@latest
openclaw gateway install --force
openclaw gateway restart
Danach TUI komplett schlieΓen und neu ΓΆffnen. Alte TUI/WebChat-Verbindungen kΓΆnnen nach Gateway-Restarts halb kaputt bleiben.
Wenn es wieder abbricht, bitte direkt danach:
openclaw gateway status
openclaw status --deep
tail -n 150 /tmp/openclaw/openclaw-2026-04-30.log
Und wichtig: poste bitte die komplette send failed: Error: ... Zeile. Gerade fehlt der eigentliche Fehlerteil nach Error:.
Nebenbei: In deinem vorherigen Status stand schon Update auf 2026.4.27 verfΓΌgbar. Da du auf 2026.4.26 bist und Gateway/WS komisch ist, wΓΌrde ich wirklich erst updaten, bevor wir tiefer debuggen.
und im tui erscheint dann kein text, ich frage was und dann kommt nichts. mehr und es wird erst gearbeitet und dann bricht es ab.
Das klingt nach: Run startet, aber liefert keinen finalen Assistant-Text, danach verliert der TUI die Verbindung oder der Send schlΓ€gt fehl. Also vermutlich Session/Provider-Run, nicht nur Anzeige.
Bitte einmal sauber isolieren:
openclaw infer model run --model openai-codex/gpt-5.5 --prompt "Sag nur: ok" --json
- Wenn das failt: Codex/Auth/Provider-Problem.
- Wenn das klappt: Codex ist okay, dann hΓ€ngt TUI/Gateway/Session-State.
Dann im TUI nicht die alte Session weiterverwenden, sondern:
/new openai-codex/gpt-5.5
Wenn es wieder abbricht, direkt wΓ€hrenddessen Logs laufen lassen:
openclaw logs --follow
und die Zeilen posten ab dem Moment, wo du die Frage abschickst β besonders error, model-fallback, incomplete turn, send failed, gateway/ws.
Und falls noch nicht gemacht: bitte auf 2026.4.27 updaten, weil dein Status schon ein Update angezeigt hat:
pnpm add -g openclaw@latest
openclaw gateway install --force
openclaw gateway restart
major@localhost-live:~$ openclaw update
Updating OpenClaw...
β
β β Updating via package manager (31.7s)
β
β β Running doctor checks (152.56s)
Update Result: OK
Root: /home/major/.npm-global/lib/node_modules/openclaw
Before: 2026.4.26
After: 2026.4.27
Total time: 187.07s
Updating plugins...
No plugin updates needed.
Completion cache update failed: Error: spawnSync /usr/bin/node ETIMEDOUT
Restarting service...
Failed to refresh gateway service environment from updated install: Error: updated install refresh failed (/home/major/.npm-global/lib/node_modules/openclaw/dist/index.js): command returned a non-zero exit code
major@localhost-live:~$ openclaw --version
OpenClaw 2026.4.27 (cbc2ba0)
major@localhost-live:~$ openclaw gateway restart
Update selbst ist durch: 2026.4.27 ist installiert β
Problem ist nur danach:
Completion cache update failed ... ETIMEDOUT
Failed to refresh gateway service environment ...
Das heiΓt: Install ok, aber der Gateway-Service wurde evtl. nicht sauber auf die neue Installation refreshed.
Wenn openclaw gateway restart hΓ€ngt oder komisch endet, mach bitte:
openclaw gateway install --force
systemctl --user daemon-reload
systemctl --user restart openclaw-gateway
sleep 10
openclaw gateway status
openclaw status --all
Wenn gateway status dann Connectivity probe: ok zeigt, teste Codex direkt:
openclaw infer model run --model openai-codex/gpt-5.5 --prompt "Sag nur: ok" --json
Wenn der Port wieder nicht lauscht oder TUI weiter abbricht, poste bitte:
journalctl --user -u openclaw-gateway.service -n 200 --no-pager
tail -n 200 /tmp/openclaw/openclaw-2026-04-30.log
major@localhost-live:~$ openclaw tui
π¦ OpenClaw 2026.4.27 (cbc2ba0) β Alexa, but with taste.
openclaw tui - ws://127.0.0.1:18789 - agent main - session main
connecting | idle
agent main | session main | unknown | tokens ?
Der TUI bleibt bei connecting hΓ€ngen β also Gateway antwortet weiterhin nicht sauber auf connect.
Bitte nicht im TUI weiter probieren, sondern Status prΓΌfen:
openclaw gateway status
openclaw status --all
Wenn da nicht Connectivity probe: ok steht, dann:
openclaw gateway install --force
systemctl --user daemon-reload
systemctl --user restart openclaw-gateway
sleep 15
openclaw gateway status
Wenn danach immer noch connecting/timeout:
journalctl --user -u openclaw-gateway.service -n 200 --no-pager
tail -n 200 /tmp/openclaw/openclaw-2026-04-30.log
Der TUI-Auszug allein sagt nur: CLI ist 2026.4.27, aber der Gateway-WS-Handshake kommt nicht fertig.
π¦ OpenClaw 2026.4.27 (cbc2ba0) β I can run local, remote, or purely on vibesβresults may vary with DNS.
β
β
Service: systemd (enabled)
File logs: /tmp/openclaw/openclaw-2026-04-30.log
Command: /usr/bin/node /home/major/.npm-global/lib/node_modules/openclaw/dist/index.js gateway --port 18789
Service file: ~/.config/systemd/user/openclaw-gateway.service
Service env: OPENCLAW_GATEWAY_PORT=18789
Service config looks out of date or non-standard.
Service config issue: Gateway service PATH includes version managers or package managers; recommend a minimal PATH. (/home/major/.volta/bin, /home/major/.asdf/shims, /home/major/.nvm/current/bin, /home/major/.fnm/aliases/default/bin, /home/major/.fnm/current/bin, /home/major/.local/share/pnpm)
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: running (pid 3069823, state active, sub running, last exit 0, reason 0)
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 3069823 major: /usr/bin/node /home/major/.npm-global/lib/node_modules/openclaw/dist/index.js gateway --port 18789 ([::1]:18789)
- Gateway already running locally. Stop it (openclaw gateway stop) or use a different port.
Listening: [::1]:18789
Troubles: run openclaw status
Troubleshooting: https://docs.openclaw.ai/troubleshooting
Gut, jetzt ist es klarer: der Port lauscht, aber der Gateway beantwortet den WS-Connect nicht β also Gateway hΓ€ngt/blocked nach dem Start. Dazu kommt:
Service config looks out of date or non-standard
PATH includes version managers/package managers
Mach als nΓ€chsten Schritt genau das:
openclaw doctor --repair
openclaw gateway install --force
systemctl --user daemon-reload
systemctl --user restart openclaw-gateway
sleep 20
openclaw gateway status
WΓ€hrenddessen TUI bitte schlieΓen, sonst erzeugt es weitere kaputte Connect-Versuche.
Wenn danach weiter Connectivity probe: failed ... timeout kommt, brauchen wir die echten Start-/Blocker-Logs:
journalctl --user -u openclaw-gateway.service -n 200 --no-pager
tail -n 200 /tmp/openclaw/openclaw-2026-04-30.log
Der Hinweis mit PATH ist wichtig β dein Update hatte ja schon Failed to refresh gateway service environment, also reparieren wir jetzt zuerst den systemd-Service sauber.
OpenClaw 2026.4.27 (cbc2ba0) β Say "stop" and I'll stopβsay "ship" and we'll both learn a lesson.
β
β
Service: systemd (enabled)
File logs: /tmp/openclaw/openclaw-2026-04-30.log
Command: /usr/bin/node /home/major/.npm-global/lib/node_modules/openclaw/dist/index.js gateway --port 18789
Service file: ~/.config/systemd/user/openclaw-gateway.service
Service env: OPENCLAW_GATEWAY_PORT=18789
Service config looks out of date or non-standard.
Service config issue: Gateway service PATH includes version managers or package managers; recommend a minimal PATH. (/home/major/.volta/bin, /home/major/.asdf/shims, /home/major/.nvm/current/bin, /home/major/.fnm/aliases/default/bin, /home/major/.fnm/current/bin, /home/major/.local/share/pnpm)
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: running (pid 3070834, state active, sub running, last exit 0, reason 0)
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 3070834 major: /usr/bin/node /home/major/.npm-global/lib/node_modules/openclaw/dist/index.js gateway --port 18789 ([::1]:18789)
- Gateway already running locally. Stop it (openclaw gateway stop) or use a different port.
Listening: [::1]:18789
Troubles: run openclaw status
Troubleshooting: https://docs.openclaw.ai/troubleshooting
Das ist leider weiterhin derselbe Zustand: Gateway-Prozess lΓ€uft und Port ist belegt, aber WS-Connect timed out. gateway status bringt uns ab jetzt nicht weiter.
Bitte jetzt nicht nochmal status posten, sondern diese drei Sachen:
openclaw doctor --repair
journalctl --user -u openclaw-gateway.service -n 200 --no-pager
tail -n 200 /tmp/openclaw/openclaw-2026-04-30.log
Wenn doctor --repair hΓ€ngt oder Fehler ausgibt, genau den Output posten.
Der wichtige Hinweis ist weiterhin:
Service config looks out of date or non-standard
PATH includes version managers/package managers
Also entweder repariert doctor --repair den systemd-Service, oder in den Logs steht, warum der Gateway nach dem Start blockiert. Ohne Journal/Gateway-Log raten wir nur weiter im Kreis.
Das Problem wenn ich mir das log anschaue dann stehen da unfassbar viele zeilen drinne
Ja, verstΓ€ndlich β dann filter es runter. Bitte diese kurzen Commands:
grep -Ei "error|warn|failed|timeout|exception|fatal|chokidar|plugin-runtime|starting HTTP|ready|agent model|SIGTERM" /tmp/openclaw/openclaw-2026-04-30.log | tail -n 120
Und fΓΌrs systemd-Journal:
journalctl --user -u openclaw-gateway.service -n 300 --no-pager | grep -Ei "error|warn|failed|timeout|exception|fatal|chokidar|plugin-runtime|starting HTTP|ready|agent model|SIGTERM" | tail -n 120
Wenn das immer noch zu viel ist, nimm tail -n 40 statt 120.
ZusΓ€tzlich einmal das hier, weil es wahrscheinlich der direkte Fix ist:
openclaw doctor --repair
Poste dann nur die Fehler/Warnungen daraus.
Immer noch zu viel
Okay, dann machen wirβs gezielt klein β nur die letzten 30 relevanten Zeilen:
grep -Ei "error|failed|fatal|exception|timeout|chokidar|plugin-runtime|received SIGTERM|ready|starting HTTP" /tmp/openclaw/openclaw-2026-04-30.log | tail -n 30
Und wenn sogar das zu viel ist:
tail -n 80 /tmp/openclaw/openclaw-2026-04-30.log
Alternativ poste nur den Abschnitt ab dem letzten:
loading configurationβ¦
bis entweder:
ready
oder bis zur ersten error/warn.
Falls du lokal schneller eingrenzen willst, probier:
openclaw doctor --repair
und poste nur, was dort mit !, β, failed oder error markiert ist.
Es sind so viele zeichen dass ich loading configuartion nicht erkennen kann
Alles gut β lass uns gar nicht mehr manuell im Log scrollen.
FΓΌhr bitte genau das aus, das erstellt dir eine kleine Datei mit nur den relevanten Stellen:
grep -Ei "loading configuration|resolving authentication|starting HTTP|ready|error|failed|fatal|exception|timeout|chokidar|plugin-runtime|SIGTERM" /tmp/openclaw/openclaw-2026-04-30.log | tail -n 50 > /tmp/openclaw-small.log
cat /tmp/openclaw-small.log
Wenn das immer noch zu viel ist:
tail -n 20 /tmp/openclaw-small.log
Und bitte auch einmal:
openclaw doctor --repair
Da brauchst du nicht alles zu posten β nur wenn es am Ende Fehler/Warnungen meldet.
Das sind 50000 zeichen, die kann ich hier unmΓΆglich posten
Nicht posten β 50k Zeichen sind Quatsch π
Bitte lass dir eine Mini-Diagnose erzeugen:
python3 - <<'PY'
from pathlib import Path
p = Path("/tmp/openclaw/openclaw-2026-04-30.log")
lines = p.read_text(errors="replace").splitlines()
keys = [
"loading configuration", "resolving authentication", "starting HTTP",
"ready", "agent model", "error", "failed", "fatal", "exception",
"timeout", "chokidar", "plugin-runtime", "SIGTERM"
]
start = max([i for i,l in enumerate(lines) if "loading configuration" in l] or [max(0, len(lines)-500)])
out = []
for n, l in enumerate(lines[start:], start + 1):
if any(k.lower() in l.lower() for k in keys):
if len(l) > 220:
l = l[:220] + " β¦"
out.append(f"{n}: {l}")
print("\n".join(out[-25:]) or "NO_MATCHES")
PY
Das sollte nur ~25 kurze Zeilen ausgeben. Nur das posten.
Und separat, falls doctor zu lang ist:
openclaw doctor --repair 2>&1 | tail -n 60
Auch davon nur die letzten 60 Zeilen.
1091: {"0":"{"subsystem":"gateway"}","1":"loading configurationβ¦","_meta":{"runtime":"node","runtimeVersion":"22.22.0","hostname":"localhost-live","name":"{"subsystem":"gateway"}","parentNames":["openclaw"],"date":"202 β¦
1092: {"0":"{"subsystem":"gateway"}","1":"resolving authenticationβ¦","_meta":{"runtime":"node","runtimeVersion":"22.22.0","hostname":"localhost-live","name":"{"subsystem":"gateway"}","parentNames":["openclaw"],"date":" β¦
1094: {"0":"{"subsystem":"gateway"}","1":"starting HTTP server...","_meta":{"runtime":"node","runtimeVersion":"22.22.0","hostname":"localhost-live","name":"{"subsystem":"gateway"}","parentNames":["openclaw"],"date":"20 β¦
1097: {"0":"{"subsystem":"gateway"}","1":"agent model: openai-codex/gpt-5.5","_meta":{"runtime":"node","runtimeVersion":"22.22.0","hostname":"localhost-live","name":"{"subsystem":"gateway"}","parentNames":["openclaw"], β¦
1103: {"0":"{"subsystem":"plugins"}","1":"plugin service failed (browser-control, plugin=browser, root=/home/major/.npm-global/lib/node_modules/openclaw/dist/extensions/browser): Cannot find package 'ajv' imported from /ho β¦
1112: {"0":"{"subsystem":"gateway"}","1":"gateway ready","_meta":{"runtime":"node","runtimeVersion":"22.22.0","hostname":"localhost-live","name":"{"subsystem":"gateway"}","parentNames":["openclaw"],"date":"2026-04-30T0 β¦
1114: {"0":"{"subsystem":"telegram/network"}","1":"fetch fallback: enabling sticky IPv4-only dispatcher (codes=none, reason=request-timeout)","_meta":{"runtime":"node","runtimeVersion":"22.22.0","hostname":"localhost-live" β¦
1117: {"0":"{"subsystem":"memory"}","1":"qmd memory unavailable; falling back to builtin: Cannot find package 'chokidar' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.27-de49e110caae/dist/qmd-mana β¦
1118: {"0":"{"subsystem":"memory"}","1":"sqlite-vec unavailable: Cannot find package 'sqlite-vec' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.27-de49e110caae/dist/engine-storage-CO3qFXo7.js\nDid β¦
1119: {"0":"{"subsystem":"gateway/channels/telegram"}","1":"telegram deleteMyCommands failed: Network request for 'deleteMyCommands' failed!","_meta":{"runtime":"node","runtimeVersion":"22.22.0","hostname":"localhost-live" β¦
1121: {"0":"{"subsystem":"gateway/channels/telegram"}","1":"telegram deleteWebhook failed: Network request for 'deleteWebhook' failed!","_meta":{"runtime":"node","runtimeVersion":"22.22.0","hostname":"localhost-live","name β¦
1122: {"0":"{"subsystem":"gateway/channels/telegram"}","1":"Telegram webhook cleanup failed: Network request for 'deleteWebhook' failed!; retrying in 2.47s.","_meta":{"runtime":"node","runtimeVersion":"22.22.0","hostname": β¦
1132: {"0":"{"subsystem":"memory"}","1":"sqlite-vec unavailable: Cannot find package 'sqlite-vec' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.27-de49e110caae/dist/engine-storage-CO3qFXo7.js\nDid β¦
1133: {"0":"{"subsystem":"memory"}","1":"chunks_vec not updated β sqlite-vec unavailable: Cannot find package 'sqlite-vec' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.27-de49e110caae/dist/engine β¦
1149: {"0":"{"subsystem":"gateway"}","1":"qmd memory startup boot sync failed for agent "main": Error: gemini embeddings failed (429): You exceeded your current quota, please check your plan and billing details. For more β¦
Aha, das ist jetzt endlich klein genug β und ziemlich eindeutig.
Gute Nachricht: Codex 5.5 ist aktiv:
agent model: openai-codex/gpt-5.5
gateway ready
Das eigentliche Problem ist daneben: kaputte/fehlende Runtime-Deps + Memory/Embedding-Fehler:
Cannot find package 'ajv'
Cannot find package 'chokidar'
Cannot find package 'sqlite-vec'
gemini embeddings failed (429)
Mach jetzt bitte:
openclaw doctor --repair --force
openclaw gateway install --force
systemctl --user daemon-reload
systemctl --user restart openclaw-gateway
sleep 20
openclaw gateway status
doctor --repair ist genau fΓΌr solche Repairs gedacht; die Docs sagen auch explizit, dass Doctor fehlende bundled plugin runtime deps repariert. --force passt hier, weil dein Service als βout of date/non-standardβ gemeldet wurde. (docs.openclaw.ai)
Falls danach noch ajv/chokidar/sqlite-vec fehlt, dann cache sauber neu aufbauen:
mv ~/.openclaw/plugin-runtime-deps ~/.openclaw/plugin-runtime-deps.bak.$(date +%s)
openclaw doctor --repair
systemctl --user restart openclaw-gateway
sleep 20
openclaw gateway status
Der Gemini-429 ist nicht Codex, sondern Memory/Embeddings. OpenClaw auto-detectet Gemini fΓΌr Memory Search, wenn ein Gemini-Key vorhanden ist; Codex OAuth deckt Embeddings nicht ab. Wenn die Quote leer ist, kann Memory-Indexing nerven/fehlschlagen. (docs.openclaw.ai)
Wenn du erstmal nur StabilitΓ€t willst, kΓΆnnen wir danach Memory Search temporΓ€r abschalten. Aber zuerst die fehlenden Runtime-Deps reparieren β das ist der dickere Fehler.
Ok ich habe das hier gemacht:
openclaw doctor --repair --force
openclaw gateway install --force
systemctl --user daemon-reload
systemctl --user restart openclaw-gateway
sleep 20
openclaw gateway status
Was mache ich jetzt?
Jetzt prΓΌfen wir, ob es wirklich sauber ist:
openclaw gateway status
openclaw status --all
Wenn dort Connectivity probe: ok steht, dann Codex direkt testen:
openclaw infer model run --model openai-codex/gpt-5.5 --prompt "Sag nur: ok" --json
Wenn das ok liefert: TUI neu starten und neue Session benutzen:
openclaw tui
im TUI dann:
/new openai-codex/gpt-5.5
Wenn gateway status weiterhin timeout zeigt oder wieder ajv/chokidar/sqlite-vec in den Logs auftaucht, dann als nΓ€chstes den Runtime-Deps-Ordner wegschieben und neu reparieren lassen:
mv ~/.openclaw/plugin-runtime-deps ~/.openclaw/plugin-runtime-deps.bak.$(date +%s)
openclaw doctor --repair
systemctl --user restart openclaw-gateway
sleep 20
openclaw gateway status
π¦ OpenClaw 2026.4.27 (cbc2ba0) β WhatsApp automation without the "please accept our new privacy policy".
Installed systemd service: /home/major/.config/systemd/user/openclaw-gateway.service
Previous unit backed up to: /home/major/.config/systemd/user/openclaw-gateway.service.bak
π¦ OpenClaw 2026.4.27 (cbc2ba0) β I keep secrets like a vault... unless you print them in debug logs again.
β
β
Service: systemd (enabled)
File logs: /tmp/openclaw/openclaw-2026-04-30.log
Command: /usr/bin/node /home/major/.npm-global/lib/node_modules/openclaw/dist/index.js gateway --port 18789
Service file: ~/.config/systemd/user/openclaw-gateway.service
Service env: OPENCLAW_GATEWAY_PORT=18789
Service config looks out of date or non-standard.
Service config issue: Gateway service PATH includes version managers or package managers; recommend a minimal PATH. (/home/major/.volta/bin, /home/major/.asdf/shims, /home/major/.nvm/current/bin, /home/major/.fnm/aliases/default/bin, /home/major/.fnm/current/bin, /home/major/.local/share/pnpm)
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: running (pid 3071916, state active, sub running, last exit 0, reason 0)
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 3071916 major: /usr/bin/node /home/major/.npm-global/lib/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
Okay β doctor/install hat den Service neu installiert, aber der Gateway timed weiterhin beim Connect aus. Jetzt bitte den nΓ€chsten Recovery-Schritt machen, den wir vorhin vorbereitet hatten:
mv ~/.openclaw/plugin-runtime-deps ~/.openclaw/plugin-runtime-deps.bak.$(date +%s)
openclaw doctor --repair
systemctl --user restart openclaw-gateway
sleep 30
openclaw gateway status
Wenn danach immer noch Connectivity probe: failed ... timeout, poste bitte nur diese kleine Ausgabe:
systemctl --user cat openclaw-gateway.service | grep -nE "ExecStart|Environment|PATH|WorkingDirectory"
und danach:
python3 - <<'PY'
from pathlib import Path
p=Path("/tmp/openclaw/openclaw-2026-04-30.log")
lines=p.read_text(errors="replace").splitlines()
for l in lines[-250:]:
if any(x in l.lower() for x in ["ready","error","failed","timeout","ajv","chokidar","sqlite-vec","gemini embeddings","plugin service"]):
print(l[:300])
PY
Wichtig: agent model: openai-codex/gpt-5.5 war schon bestΓ€tigt. Jetzt debuggen wir nur noch, warum der Gateway beim WebSocket-Connect hΓ€ngen bleibt.
major@localhost-live:~$ systemctl --user cat openclaw-gateway.service | grep -nE "ExecStart|Environment|PATH|WorkingDirectory"
10:ExecStart=/usr/bin/node /home/major/.npm-global/lib/node_modules/openclaw/dist/index.js gateway --port 18789
18:Environment=OPENCLAW_SERVICE_MANAGED_ENV_KEYS=GEMINI_API_KEY
19:Environment=HOME=/home/major
20:Environment=TMPDIR=/tmp
21:Environment=PATH=/usr/bin:/home/major/.local/bin:/home/major/.npm-global/bin:/home/major/bin:/home/major/.nix-profile/bin:/usr/local/bin:/bin:/home/major/.volta/bin:/home/major/.asdf/shims:/home/major/.bun/bin:/home/major/.nvm/current/bin:/home/major/.local/share/fnm/aliases/default/bin:/home/major/.local/share/fnm/current/bin:/home/major/.fnm/aliases/default/bin:/home/major/.fnm/current/bin:/home/major/.local/share/pnpm
22:Environment=OPENCLAW_GATEWAY_PORT=18789
23:Environment=OPENCLAW_SYSTEMD_UNIT=openclaw-gateway.service
24:Environment="OPENCLAW_WINDOWS_TASK_NAME=OpenClaw Gateway"
25:Environment=OPENCLAW_SERVICE_MARKER=openclaw
26:Environment=OPENCLAW_SERVICE_KIND=gateway
27:Environment=OPENCLAW_SERVICE_VERSION=2026.4.27
Exec approval is required, but no interactive approval client is currently available.
Approve it from the Web UI or terminal UI, or enable a native chat approval client such as Discord. If those accounts already know your owner ID via allowFrom or owner config, OpenClaw can often infer approvers automatically. Then retry the command. You can usually leave execApprovals.approvers unset when owner config already identifies the approvers.
Yep, das bestΓ€tigt den PATH-Warnhinweis: der Service hat immer noch Volta/asdf/nvm/fnm/pnpm im PATH.
Setz jetzt einen sauberen systemd-Override:
mkdir -p ~/.config/systemd/user/openclaw-gateway.service.d
cat > ~/.config/systemd/user/openclaw-gateway.service.d/10-clean-path.conf <<'EOF'
[Service]
Environment=PATH=/usr/bin:/usr/local/bin:/bin:/home/major/.npm-global/bin:/home/major/.local/bin
EOF
systemctl --user daemon-reload
systemctl --user restart openclaw-gateway
sleep 30
openclaw gateway status
Wenn danach weiterhin timeout kommt, dann bitte nochmal die Mini-Log-Ausgabe posten β aber erstmal nur nach dem Restart:
python3 - <<'PY'
from pathlib import Path
p=Path("/tmp/openclaw/openclaw-2026-04-30.log")
for l in p.read_text(errors="replace").splitlines()[-200:]:
if any(x in l.lower() for x in ["ready","error","failed","timeout","ajv","chokidar","sqlite-vec","gemini embeddings","plugin service"]):
print(l[:260])
PY
Und ja: Codex 5.5 ist nicht mehr das Problem β agent model: openai-codex/gpt-5.5 war bestΓ€tigt. Jetzt ist es Service/runtime stability.
Kleines Meta: mein lokaler Suchversuch in den OpenClaw-Dateien wurde wegen Approval-Timeout nicht ausgefΓΌhrt. Ist aber nicht blockierend β dein eigener Output reicht hier.
Bleib bei dem Override-Schritt von oben:
mkdir -p ~/.config/systemd/user/openclaw-gateway.service.d
cat > ~/.config/systemd/user/openclaw-gateway.service.d/10-clean-path.conf <<'EOF'
[Service]
Environment=PATH=/usr/bin:/usr/local/bin:/bin:/home/major/.npm-global/bin:/home/major/.local/bin
EOF
systemctl --user daemon-reload
systemctl --user restart openclaw-gateway
sleep 30
openclaw gateway status
Wenn der Status danach immer noch timeout zeigt, bitte die Mini-Log-Ausgabe posten.
major@localhost-live:~$ python3 - <<'PY'
from pathlib import Path
p=Path("/tmp/openclaw/openclaw-2026-04-30.log")
for l in p.read_text(errors="replace").splitlines()[-200:]:
if any(x in l.lower() for x in ["ready","error","failed","timeout","ajv","chokidar","sqlite-vec","gemini embeddings","plugin service"]):
print(l[:260])
PY
{"0":"{"subsystem":"plugins"}","1":"plugin service failed (browser-control, plugin=browser, root=/home/major/.npm-global/lib/node_modules/openclaw/dist/extensions/browser): Cannot find package 'ajv' imported from /home/major/.openclaw/plugin-runtime-deps/o
{"0":"{"subsystem":"gateway"}","1":"gateway ready","_meta":{"runtime":"node","runtimeVersion":"22.22.0","hostname":"localhost-live","name":"{"subsystem":"gateway"}","parentNames":["openclaw"],"date":"2026-04-30T00:59:09.538Z","logLevelId":3,"logLevelNa
{"0":"{"subsystem":"telegram/network"}","1":"fetch fallback: enabling sticky IPv4-only dispatcher (codes=none, reason=request-timeout)","_meta":{"runtime":"node","runtimeVersion":"22.22.0","hostname":"localhost-live","name":"{"subsystem":"telegram/netwo
{"0":"{"subsystem":"memory"}","1":"qmd memory unavailable; falling back to builtin: Cannot find package 'chokidar' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.27-de49e110caae/dist/qmd-manager-BCC9vRXM.js","_meta":{"runtime":"nod
{"0":"Installing bundled plugin runtime deps (17 missing, 19 install specs): @agentclientprotocol/claude-agent-acp@0.31.0, @grammyjs/runner@^2.0.3, @grammyjs/transformer-throttler@^1.2.1, @homebridge/ciao@^1.3.6, @mariozechner/pi-ai@0.70.5, @modelcontextprotoc
{"0":"Installed bundled plugin runtime deps in 35s: @agentclientprotocol/claude-agent-acp@0.31.0, @grammyjs/runner@^2.0.3, @grammyjs/transformer-throttler@^1.2.1, @homebridge/ciao@^1.3.6, @mariozechner/pi-ai@0.70.5, @modelcontextprotocol/sdk@1.29.0, @zed-indus
{"0":"{"subsystem":"memory"}","1":"sqlite-vec unavailable: Cannot find package 'sqlite-vec' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.27-de49e110caae/dist/engine-storage-CO3qFXo7.js\nDid you mean to import "sqlite-vec/index.c
{"0":"{"subsystem":"gateway/channels/telegram"}","1":"telegram deleteWebhook failed: Network request for 'deleteWebhook' failed!","_meta":{"runtime":"node","runtimeVersion":"22.22.0","hostname":"localhost-live","name":"{"subsystem":"gateway/channels/tel
{"0":"{"subsystem":"gateway/channels/telegram"}","1":"Telegram webhook cleanup failed: Network request for 'deleteWebhook' failed!; retrying in 2.48s.","_meta":{"runtime":"node","runtimeVersion":"22.22.0","hostname":"localhost-live","name":"{"subsystem":
{"0":"{"subsystem":"gateway/channels/telegram"}","1":"telegram deleteMyCommands failed: Network request for 'deleteMyCommands' failed!","_meta":{"runtime":"node","runtimeVersion":"22.22.0","hostname":"localhost-live","name":"{"subsystem":"gateway/channe
{"0":"{"subsystem":"memory"}","1":"sqlite-vec unavailable: Cannot find package 'sqlite-vec' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.27-de49e110caae/dist/engine-storage-CO3qFXo7.js\nDid you mean to import "sqlite-vec/index.c
{"0":"{"subsystem":"memory"}","1":"chunks_vec not updated β sqlite-vec unavailable: Cannot find package 'sqlite-vec' imported from /home/major/.openclaw/plugin-runtime-deps/openclaw-2026.4.27-de49e110caae/dist/engine-storage-CO3qFXo7.js\nDid you mean to im
{"0":"{"subsystem":"gateway"}","1":"[plugins] staging bundled runtime deps before gateway startup (5 missing, 18 install specs): @agentclientprotocol/claude-agent-acp@0.31.0, @homebridge/ciao@^1.3.6, @zed-industries/codex-acp@0.12.0, acpx@0.6.1, chokidar@^
{"0":"{"subsystem":"gateway"}","1":"[plugins] installed bundled runtime deps before gateway startup in 6961ms: @agentclientprotocol/claude-agent-acp@0.31.0, @homebridge/ciao@^1.3.6, @zed-industries/codex-acp@0.12.0, acpx@0.6.1, chokidar@^5.0.0","_meta":{"r
Das sieht nach alten Fehlern + danach Repair aus. Die wichtigen letzten Zeilen sind gut:
[plugins] staging bundled runtime deps before gateway startup ...
[plugins] installed bundled runtime deps ...
HeiΓt: er hat fehlende Runtime-Deps jetzt nachinstalliert. Jetzt einmal frisch nach dem Repair testen:
systemctl --user restart openclaw-gateway
sleep 45
openclaw gateway status
Wenn dort Connectivity probe: ok steht:
openclaw infer model run --model openai-codex/gpt-5.5 --prompt "Sag nur: ok" --json
Wenn es immer noch timeout ist, wΓΌrde ich als NΓ€chstes Memory Search temporΓ€r abschalten, weil dein Log auch Gemini-Embedding-429 und sqlite-vec/QMD-Probleme zeigt:
openclaw config set agents.defaults.memorySearch.enabled false
systemctl --user restart openclaw-gateway
sleep 30
openclaw gateway status
Das ist reversibel. Codex bleibt davon unberΓΌhrt β Codex 5.5 war ja schon aktiv.
Es hat funktioniert nach dem letzten snippet ich bin dann wieder zu codex ΓΌber telegram hingegangen und habe versucht, es zu fixen, aber jetzt meldet sich dieser nicht mehr:
[30.04.26 04:34] Zeus: Ich schau kurz in Status + Logs, dann sag ich dir den konkreten Fix.
[30.04.26 04:34] Zeus: Jetzt ist es ok:
β’ Connectivity probe: ok
β’ Capability: admin-capable
β’ Gateway lΓ€uft auf 127.0.0.1:18789
Ursache: Der Gateway war nach sleep 45 noch nicht wirklich ready. In den Logs: Start dauerte ca. 90s+ wegen Plugin/Runtime-Dep-Setup, Telegram-Network-Timeouts und Model-Warmup. Deshalb kam erst Timeout, dann spΓ€ter ok.
Der Port 18789 is already in use ist hier kein Fehler β das ist genau der laufende Gateway-Prozess.
Separat bleibt nur die Warnung:
Service config PATH includes version managers/package managers
DafΓΌr spΓ€ter mal sauber:
openclaw doctor --repair
systemctl --user restart openclaw-gateway
Kurz: nicht kaputt, nur zu frΓΌh geprΓΌft π
[30.04.26 04:41] Tobias: OK, also ist soweit alles in Ordnung?
[30.04.26 04:41] Zeus: Ja β soweit alles in Ordnung. β
Der Gateway lΓ€uft, WebSocket-Probe ist ok, Dashboard erreichbar, Capability admin-capable.
Einzige Baustelle ist kosmetisch/Hardening:
openclaw doctor --repair
wegen der PATH-Warnung. Muss nicht sofort sein.
[30.04.26 04:42] Tobias: Was bedeutet die warnung genau?
[30.04.26 04:42] Zeus: Die Warnung heiΓt:
Dein openclaw-gateway.service startet mit einem zu groΓen/βschmutzigenβ PATH, der u.a. Version-Manager enthΓ€lt:
β’ Volta
β’ asdf
β’ nvm/fnm
β’ pnpm
β’ evtl. lokale Shims
Systemd-Services sollten mΓΆglichst deterministisch starten, z.B. mit:
/usr/local/bin:/usr/bin:/bin
Warum OpenClaw meckert:
β’ Version-Manager kΓΆnnen plΓΆtzlich eine andere node/npm/Binary-Version liefern
β’ Updates/Shell-Profile kΓΆnnen den Service unbemerkt verΓ€ndern
β’ Debugging wird schwerer: Terminal β systemd-Umgebung
β’ im Worst Case startet Gateway nach Reboot anders als jetzt
Bei dir ist es aber nicht kritisch, weil die echte Command-Zeile schon sauber aussieht:
/usr/bin/node /home/major/.npm-global/lib/node_modules/openclaw/dist/index.js gateway --port 18789
Also: Gateway lΓ€uft korrekt. Die Warnung sagt nur: βMach den Service-PATH minimalistischer, damit es robuster ist.β