openclaw - 💡(How to fix) Fix [Bug]: OpenClaw Bug Report: Stuck Session CPU Regression in v2026.4.29 [2 comments, 3 participants]

Bug type

Regression (worked before, now fails)

Beta release blocker

No

Summary

After updating to OpenClaw v2026.4.29, the gateway process consumes 47-100% CPU continuously, causing severe event loop delays (up to 22 seconds) and breaking Discord integration.

Steps to reproduce

1. Update to OpenClaw v2026.4.29
2. Have Discord channel sessions active
3. Observe CPU spike and stuck session warnings in logs
4. Discord inbound stops working

Expected behavior

• CPU usage: <10% idle
• Event loop delay: <100ms
• Discord: Messages work bidirectionally
• Sessions: Clean up properly after processing

Actual behavior

• CPU: 47-100% continuous
• Event loop: 2-22 second delays
• Discord: Outbound works, inbound broken (timeouts)
• Sessions: Stuck in processing state, diagnostic loop spam

OpenClaw version

2026.4.29

Operating system

Ubuntu (VirtualBox VM)

Install method

No response

Model

.

Provider / routing chain

.

Additional provider/model setup details

No response

Logs, screenshots, and evidence

## Log Snippets

liveness warning: reasons=event_loop_delay interval=32s 
  eventLoopDelayP99Ms=16458.4 eventLoopUtilization=1 cpuCoreRatio=1.034

fetch timeout reached; aborting operation
  url=https://discord.com/api/v10/users/@me elapsedMs=20872

Impact and severity

High - Breaks Discord integration, renders system unusable on limited resources.

Additional information

Environment

• Version: 2026.4.29 (d11e02136d2b)
• OS: Ubuntu (VirtualBox VM)
• CPU: 3 cores
• RAM: 5.4 GB
• Node: v22.22.2

Regression Confirmation

• Previous version: Worked fine before 2026.4.29
• Discord worked: Yes, before update
• CPU usage: Normal before update

Symptoms

1. High CPU: 47-100% continuous usage by openclaw process
2. Event loop delays: 2-22 seconds (should be <100ms)
3. Discord broken: Inbound messages don't work, API calls timeout after 20+ seconds
4. Slow responses: All API calls 10-100x slower than normal

Root Cause Identified

Stuck session detection/recovery loop spam:

stuck session: sessionId=main sessionKey=agent:main:discord:channel:[MASKED]
  state=processing age=123s queueDepth=1 reason=processing_with_queued_work recovery=checking

stuck session recovery skipped: reason=active_embedded_run action=observe_only 
  sessionId=main age=153s queueDepth=1 activeSessionId=[MASKED]

This repeats every 30 seconds, consuming CPU cycles and filling logs.

Technical Details

• Session stuck in processing state for 200+ seconds
• Diagnostic system continuously polls stuck sessions
• File lock held: [MASKED_SESSION].jsonl.lock
• 17 threads active (reasonable, not a leak)
• 303 session files in directory (782MB total)

Thread Breakdown

• 6x tokio-runtime-w (Rust/WASM workers)
• 5x node (Node.js threads)
• 4x libuv-worker
• 1x DelayedTaskScheduler
• 1x openclaw main

Affected Channel

• Discord channel (debug-yourself) — IDs masked for privacy

Workarounds Attempted

1. ✅ Cleared session lock files
2. ✅ Gateway restart
3. ✅ Session directory cleanup
4. ✅ Downgrade to v2026.4.14 (confirmed working)

Related Issues

• Discord Gateway timeout: fetch timeout reached; aborting operation (20s)
• API calls timing out: sessions.list taking 6+ seconds