Gateway issie | Friends of the Crustacean 🦞🤝 | Page 1

ocean musk Apr 24, 2026, 12:39 AM

#

⚠️ Something went wrong while processing your request. Please try again, or use /new to start a fresh session.

After updating to 4.21, this message comes up whenever I try to talk to my bot in telegram. Strangely, the cron jobs stay and pop up in telegram as usual. How to fix this? 🙏🙏

earnest oarBOT Apr 24, 2026, 12:39 AM

#

keen ridgeBOT Apr 24, 2026, 12:39 AM

#

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

earnest oarBOT Apr 24, 2026, 12:39 AM

#

uncut shadowBOT Apr 24, 2026, 12:41 AM

#

Let me check the actual docs first so I don't guess.

#

Need one more pass through the docs/code for likely update regressions around Telegram sessions.

#

Since cron messages still arrive, Telegram itself is probably fine. This usually points to the interactive agent run/session path, not the Telegram transport.

I’d try this in order:

In Telegram, send:

/new

If that fixes it, the upgrade likely left an old session in a bad state.

On the host, run:

openclaw --version
openclaw status --all
openclaw doctor --fix
openclaw sessions cleanup --enforce --fix-missing

#

openclaw gateway restart

Then retry from Telegram. If it still fails, tail logs while sending one message:

openclaw logs --follow

Two common causes I’d specifically look for:

stale/broken session state after upgrade
Active Memory timing out before reply, which can surface exactly as this generic error

If you see active-memory: ... status=timeout in logs, temporarily disable it and test again:

#

/active-memory off

Docs:

FAQ: https://docs.openclaw.ai/help/faq
Troubleshooting: https://docs.openclaw.ai/gateway/troubleshooting
Sessions cleanup: https://docs.openclaw.ai/cli/sessions
Active Memory: https://docs.openclaw.ai/concepts/active-memory

If you paste:

openclaw --version
the output of openclaw status --all
the first error lines from openclaw logs --follow right after a failed Telegram message

…I can narrow it down pretty fast.

ocean musk Apr 24, 2026, 7:14 AM

#

ham2sexy@Mac-mini .openclaw % openclaw --version
OpenClaw 2026.4.21 (f788c88)
ham2sexy@Mac-mini .openclaw % openclaw status --all

🦞 OpenClaw 2026.4.21 (f788c88)
Powered by open source, sustained by spite and good documentation.

│
OpenClaw status --all

Overview

│ Item │ Value
│ Version │ 2026.4.21 │
│ OS
│ Node │ 22.20.0 │ Tailscale │ off default) │
│ Update │ pnpm · npm update 2026.4.22 │
│ Gateway │ local · ws://127.0.0.1:18789 (local loopback) · unreachable │
│ │ (unauthorized: gateway token mismatch (provide gateway auth │
│ │ token)) │
│ Security │ Run: openclaw security audit --deep │
│ Gateway self │ unknown │
│ Gateway service │ LaunchAgent installed · loaded · running (pid 35501, state │
│ │ active) │
│ Node service │ LaunchAgent not installed │
│ Agents │ 1 total · 1 bootstrapping · 1 active · 41 sessions │
│ Secrets │ none
Channels
│ Channel │ Enabled │ State │
│ Telegram │ ON │ OK │ token config (sha256:80359197 · len 46) · accounts │
│ │ │ │ 1/1
│ Account │ Status │ Notes
Diagnosis (read-only)

Gateway connection details:
Gateway target: ws://127.0.0.1:18789
Source: local loopback
Config: /Users/ham2sexy/.openclaw/openclaw.json
Bind: loopback

#

🦞 OpenClaw 2026.4.21 (f788c88)
I don't have opinions about tabs vs spaces. I have opinions about everything else.

gateway connect failed: GatewayClientRequestError: unauthorized: gateway token mismatch (set gateway.remote.token to match gateway.auth.token)
Gateway not reachable. Is it running and accessible?
Gateway target: ws://127.0.0.1:18789
Source: local loopback
Config: /Users/ham2sexy/.openclaw/openclaw.json
Bind: loopback
Hint: run openclaw doctor.
ham2sexy@Mac-mini .openclaw %

uncut shadowBOT Apr 24, 2026, 7:16 AM

#

Nice, your status --all already shows the main clue:

Gateway ... unauthorized: gateway token mismatch

So this looks less like a Telegram problem, and more like your CLI and the running LaunchAgent are using different gateway auth/config. That also explains why cron can still post, while normal chat runs are weird.

I’d do this first:

openclaw gateway status --deep
openclaw gateway install --force
openclaw gateway restart
openclaw status --all

Important: run that from the same shell/profile/env you want the service to use. If you use --profile or OPENCLAW_PROFILE, use it consistently there too.

#

Why: the docs call out that when Config (cli) and Config (service) drift, the fix is to reinstall the gateway service from the desired profile/env:
https://docs.openclaw.ai/help/faq#why-does-openclaw-gateway-status-show-config-cli-and-config-service-different

After that, if Telegram still throws the generic error, do:

openclaw doctor --fix
openclaw sessions cleanup --enforce --fix-missing
openclaw logs --follow

Then send one Telegram message and watch the logs.

Two extra clues from your output:

1 bootstrapping is suspicious, so I want to see whether the agent is stuck during startup
because your CLI currently cannot auth to the gateway, session cleanup / doctor may have been targeting the wrong runtime until the service drift is fixed

#

If the logs show active-memory ... timeout, try this in Telegram and test again:

/active-memory off

If you paste back:

openclaw gateway status --deep
the first error lines from openclaw logs --follow right after one failed Telegram message

…I can probably pinpoint it fast.
Docs: https://docs.openclaw.ai/gateway/troubleshooting

#

Yep, this is the real problem now: local gateway auth drift.

Your CLI is trying to connect to the local gateway with the wrong token:

unauthorized: gateway token mismatch

So I’d fix gateway auth first, then retest Telegram.

Run these:

openclaw config get gateway.mode
openclaw config get gateway.auth.token
openclaw config get gateway.remote.token

If gateway.mode is local, then gateway.remote.token is usually just stale noise. Remove it:

#

openclaw config unset gateway.remote.token
openclaw config unset gateway.remote.password

Then reinstall the service so the LaunchAgent and CLI use the same config/env again:

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

If gateway.auth.token is empty/missing, generate one first:

#

openclaw doctor --generate-gateway-token
openclaw gateway install --force
openclaw gateway restart

What I expect after that:

openclaw status --all should stop showing gateway token mismatch
then Telegram chat replies should usually start working again

If it still fails after the token mismatch is gone, then do:

openclaw logs --follow

and send one Telegram message. At that point we’ll be looking at the actual reply error instead of the gateway auth drift.

#Gateway issie