openclaw - 💡(How to fix) Fix Gateway health reports event_loop_utilization degraded but p99/max=0ms and actual CPU ~2% [1 comments, 2 participants]

Observation

Since upgrading to 2026.5.7, openclaw health persistently reports:

Gateway event loop: degraded reasons=event_loop_utilization,cpu max=0ms p99=0ms util=1 cpu=0.969

Evidence this is a metric reporting bug

• max=0ms and p99=0ms: a truly pegged event loop would show nonzero latency
• Actual process CPU via ps aux: ~2-6% (not ~97%)
• Gateway responds normally; no timeout or slowness observed
• No journal warnings

Environment

• Version: 2026.5.7 (upgraded from 2026.5.3-1 on 2026-05-09)
• Platform: Ubuntu 24.04, Node.js via systemd user service
• Persisted 24h+ across multiple gateway restarts

Ask

Clarify what util=1 means in the health output — is this a normalized 0-1 ratio being misread as a percentage, or is the degraded threshold too sensitive? The contradictory latency metrics suggest the degraded state is a false positive.