#WhatsApp connector flapping — root cause was mDNS, not WhatsApp

1 messages · Page 1 of 1 (latest)

keen torrent
#

Hey 👋 Posting in case this helps someone else (or if it's a known bug worth filing).

Symptom: WhatsApp channel stuck cycling linked → running → disconnected roughly every 30s. Looked like a WhatsApp/Baileys problem at first glance.

Actual cause: the gateway itself was crashing every ~30s on unhandled promise rejections from the bonjour/CIAO mDNS advertiser:

[plugins] bonjour: restarting advertiser (service stuck in announcing for 15s ...)
[openclaw] Unhandled promise rejection: CIAO ANNOUNCEMENT CANCELLED

Every gateway restart kicked WhatsApp back through the connect dance, hence the flapping.

Workaround that fixed it:

openclaw config set discovery.mdns.mode off
openclaw gateway restart

(We don't need wide-area mDNS discovery on a loopback gateway anyway.)

Question for the maintainers: should the CIAO advertiser catch/swallow ANNOUNCEMENT CANCELLED rejections instead of letting them bubble up and take the gateway down? Happy to file an issue on GitHub if useful — just wanted to flag here first.

Env: macOS (Darwin 25.4.0 arm64), node v25.9.0, OpenClaw via Homebrew.

uncut oracleBOT
#

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

frozen gladeBOT
#

Nice catch — yep, that should not take the gateway down.

I checked current source/changelog: recent OpenClaw already has Bonjour/CIAO handling for both CIAO ANNOUNCEMENT CANCELLED and CIAO PROBING CANCELLED, so:

  1. Please check:
    openclaw --version
  2. If you’re below recent 2026.4.25/2026.4.26 builds, update first.
  3. If this still reproduces on latest Homebrew OpenClaw, please file it here with your log snippet + openclaw status --all:
    https://github.com/openclaw/openclaw/issues

Your workaround is valid for now:

openclaw config set discovery.mdns.mode off