Summary

Telegram message tool with action=send fails with "Network request for 'sendMessage' failed!" while normal conversation replies work fine. The gateway doesn't crash but all proactive sends fail.

Environment

• OS: Linux 6.8.0-90-generic (x64)
• Node: v22.22.0
• OpenClaw: 2026.1.29
• Channel: Telegram (grammY)
• Gateway: Local (loopback)

Steps to Reproduce

1. Configure Telegram channel with bot token
2. Start gateway normally
3. Receive messages from user (works ✅)
4. Send normal replies (works ✅)
5. Use message tool with action=send to send proactive message
6. Observe failure: Network request for 'sendMessage' failed!

Observed Behavior

[telegram] message failed: Network request for 'sendMessage' failed!
[tools] message failed: Network request for 'sendMessage' failed!
[openclaw] Non-fatal unhandled rejection (continuing): TypeError: fetch failed
    at node:internal/deps/undici/undici:14902:13
    at processTicksAndRejections (node:internal/process/task_queues:105:5)

Key Observations

• ✅ Gateway receives incoming Telegram messages
• ✅ Normal agent replies (via session) work perfectly
• ❌ Tool-based sends (message action=send) fail consistently
• ❌ sendChatAction also fails with same error
• Gateway log shows: autoSelectFamily=false (default-node22)
• Manual curl to Telegram API works
• Manual node -e "fetch(...)" in new process works
• Only the gateway process has the issue

Logs

Gateway startup:

2026-01-31T09:47:28.558Z [telegram] autoSelectFamily=false (default-node22)

Repeated failures:

2026-01-31T09:47:44.211Z [telegram] message failed: Network request for 'sendMessage' failed!
2026-01-31T09:47:44.248Z [openclaw] Non-fatal unhandled rejection (continuing): TypeError: fetch failed
2026-01-31T09:47:45.219Z [telegram] sendChatAction failed: Network request for 'sendChatAction' failed!

Expected Behavior

The message tool should successfully send messages to Telegram, just like normal conversation replies do.

Workaround Attempted

• ✅ Verified bot token is valid (manual API calls work)
• ✅ Restarted gateway with gateway.restart
• ✅ Full systemctl stop/start
• ❌ Problem persists after all restarts

Hypothesis

Possible IPv4/IPv6 dual-stack issue with Node 22's autoSelectFamily=false setting, or network state corruption in the gateway process that doesn't affect normal reply flow.

Related Issues

• Telegram message delivery silently fails on fetch error (no retry) #3072 (Telegram fetch failures, recent comment mentions IPv6)
• Gateway crashes on unhandled fetch failure (TypeError: fetch failed) #5152 (fetch failed causing crashes, though gateway doesn't crash in this case)
• Unhandled fetch rejections crash the gateway #5172, Unhandled fetch failures cause gateway crash #5199 (unhandled fetch rejections)