Description

A significant performance regression has been observed when upgrading from v2026.4.23 to later versions (reproduced on v2026.4.27 and v2026.4.29). The system becomes unresponsive, with commands like openclaw status taking a long time to return, and background tasks (like Feishu/Lark integration) hitting 10s timeouts.

Environment

• OS: Rocky Linux 9 (Kernel 5.14.0-611.47.1.el9_7.x86_64)
• Deployment: Docker (v27.5.0) with custom image build
• Node.js: v24.14.0 (inside container)
• OpenClaw Version:
  • Stable: v2026.4.23
  • Regression observed: v2026.4.27, v2026.4.29

Diagnostic Logs (from problematic versions)

In the problematic versions, the following warnings/errors are frequent:

1. High Event Loop Delay:
   liveness warning: reasons=event_loop_delay,event_loop_utilization,cpu interval=32s eventLoopDelayP99Ms=9042.9 eventLoopDelayMaxMs=9042.9 eventLoopUtilization=1 cpuCoreRatio=1.095
2. Network Timeouts:
   AxiosError: timeout of 10000ms exceeded during Feishu/Lark bot ping.
3. Resource Errors:
   [channels] failed to load bundled channel irc: ENOENT: no such file or directory, open '/var/lib/openclaw/plugin-runtime-deps/.../dist/chrome-BTePW5f5.js'

Steps to Reproduce

1. Deploy OpenClaw v2026.4.23 -> Performance is excellent, openclaw status is instant.
2. Upgrade to v2026.4.27 or v2026.4.29 (git checkout & rebuild).
3. Observe high CPU utilization and extreme lag (several seconds) for any CLI command.

Expected Behavior

Performance should be as responsive as in v2026.4.23.

Actual Behavior

The main event loop is blocked for ~9 seconds, likely due to heavy synchronous operations or plugin scanning logic introduced after v2026.4.23.