Bug

Version: 2026.5.2 (also affects 2026.4.29+)
OS: Windows 10/11 (x64), NT 10.0.26200
Node: v24.14.0
Channel: Telegram

Description

After upgrading to 2026.5.2, all inbound Telegram media fail — both audio and images.

The bot replies: Failed to download media. Please try again.

Root Cause

The Node.js event loop is completely saturated:

• eventLoopDelayP99Ms = 11064ms (should be <50ms)
• eventLoopUtilization = 0.999 (99.9%)
• cpuCoreRatio = 1.034
• Fetch timeout on api.telegram.org/getMe after 10000ms+ (actual 12775ms)

All async operations including media downloads time out or fail.

Logs

[diagnostic] liveness warning: reasons=event_loop_delay,event_loop_utilization,cpu interval=31s eventLoopDelayP99Ms=11064.6 eventLoopDelayMaxMs=11064.6 eventLoopUtilization=0.999 cpuCoreRatio=1.034 active=1 waiting=0 queued=1
[fetch-timeout] fetch timeout after 10000ms (elapsed 12775ms) operation=fetchWithTimeout url=https://api.telegram.org/bot.../getMe

Workaround

Downgrade to 2026.4.23 fixes everything:

• Media downloads work normally (audio + images)
• Event loop is healthy
• No fetch timeouts

Environment

• Node v24.14.0
• Windows NT 10.0.26200 (x64)
• No proxy/VPN interfering
• DNS resolves correctly for api.telegram.org (IPv4 + IPv6 both present)
• HTTP 302 from api.telegram.org confirmed via curl
• npm global install, no Docker

Related

Same regression pattern as previously reported in #75768 (latency regression on Windows from 2026.4.29).

The 2026.4.21 version was the last known-good version before this class of issues appeared.