Running v2026.3.1 on OpenClawCloud hosted instance. Telegram stopped working after Tailscale was configured and now won’t recover. Gateway is running, web chat works fine, but all Telegram API calls fail with “Network request failed” on deleteWebhook and setMyCommands.
Tried the documented IPv6 fix (channels.telegram.networkOptions.family: 4) but OpenClawCloud’s config rejects it as an unrecognized key — so the fix is blocked at the infrastructure level.
This matches GitHub issues #21043 and #4847. The server appears to be attempting IPv6 connections to Telegram’s API that are unreachable on your VM infrastructure.
Two questions:
1. Can you apply the IPv4 forcing fix at the infrastructure level?
2. Can you credit the API credits burned during the heartbeat loop while Telegram was broken?
Happy to share logs if needed.
#Telegram broken on OpenClawCloud — IPv6/network issue, not config
1 messages · Page 1 of 1 (latest)
You’re not wrong — this does match the Telegram IPv6/egress failure pattern.
Quick correction first:
channels.telegram.networkOptions.family is not a current config key.
The supported path is under channels.telegram.network:
channels: {
telegram: {
network: {
autoSelectFamily: false,
dnsResultOrder: "ipv4first",
},
},
}
Docs:
- https://docs.openclaw.ai/channels/telegram
- https://docs.openclaw.ai/help/faq#telegram-setmycommands-fails-what-should-i-check
For your 2 questions:
- Infra-level IPv4 forcing: needs cloud/maintainer-side action (not something I can apply from this thread).
- Credit reimbursement: also needs maintainer/billing action.
Best next step: open a GitHub issue (or comment on #21043) with:
openclaw --version- exact config validation error text
- log lines showing
deleteWebhook/setMyCommandsnetwork failures - instance/region details (if safe to share)
- explicit reimbursement request for heartbeat burn during outage
Issue tracker: https://github.com/openclaw/openclaw/issues