Bug type

Regression (worked before, now fails)

Beta release blocker

No

Summary

Environment:

• OpenClaw 2026.4.5
• macOS arm64 (Mac mini)
• Two Telegram bot accounts configured
• Node 24.14.1

Problem: The Telegram polling runner intermittently stalls with Network request for 'getUpdates' failed! errors. This causes:

• Delayed inbound message delivery (minutes to hours)
• Missed heartbeats and cron job delivery failures
• Group chat messages never received (both bot accounts affected)

Steps to reproduce

1. Install OpenClaw 2026.4.5 on macOS arm64 with two Telegram bot accounts configured.
2. Run openclaw gateway start — both bots begin polling.
3. Wait approximately 30–90 minutes (or overnight).
4. Observe: polling stalls with "Network request for 'getUpdates' failed!" errors begin occurring on both accounts simultaneously.
5. Send a Telegram DM or group message to the bot — no response received during stall.
6. Run openclaw gateway restart — polling recovers temporarily, then stalls again within minutes to hours.
7. Note: curl https://api.telegram.org/bot<token>/getMe succeeds from the same machine throughout, confirming the network itself is not the issue.

Expected behavior

Rapid response in Telegram

Actual behavior

Telegram polling runner stalls intermittently with "Network request for 'getUpdates' failed!" errors. During stalls (lasting minutes to hours), inbound messages are not received and the agent is unresponsive. Gateway logs show repeated cycles of: stall detected → forced restart → rebuild transport → stall again. Group chat messages are never delivered regardless of polling state. A gateway restart temporarily resolves DM delivery but the stall recurs.

OpenClaw version

2026.4.5

Operating system

macOS arm64 (Mac mini)

Install method

Change to npm global install (sudo npm install -g openclaw)

Model

Change to anthropic/claude-sonnet-4-6

Provider / routing chain

Change to anthropic/claude-sonnet-4-6

Additional provider/model setup details

No response

Logs, screenshots, and evidence

Telegram polling runner stopped (polling stall detected); restarting in 30s
[diag] polling cycle finished reason=polling stall detected durationMs=30004 error=Network request for 'getUpdates' failed!
Polling runner stop timed out after 15s; forcing restart cycle

Impact and severity

No response

Additional information

• curl https://api.telegram.org/bot.../getMe works fine from the same machine — network is not the issue • Stalls affect both bot accounts simultaneously • A gateway restart temporarily clears the stall, but it recurs • DM delivery works when polling is healthy; group messages never work regardless • Problem persisted across update from 2026.4.2 → 2026.4.5 • Config has autoSelectFamily: false and dnsResultOrder: ipv4first (suggest this may be related)

Suspected cause: Something in the HTTP client used for long-polling getUpdates is losing its connection silently and not recovering properly. May be related to dual bot account setup or a macOS-specific networking quirk.