Summary

On a macOS host (Mac mini 2018, 8GB), the Gateway frequently takes 60–200s to respond to cron.status / cron.list, causing the OpenClaw CLI to error with the default --timeout 10000 ("Error: gateway timeout after 10000ms").

The gateway is healthy (openclaw gateway status → RPC probe: ok) and the cron store is small (~19KB, 9 jobs). This appears to be RPC latency / lane contention, not a corrupted cron store.

Environment

• OpenClaw: 2026.2.2-3
• OS: macOS Sequoia 15.7.2 (Darwin 24.6.0)
• Node: 24.12.0 (service command: /usr/local/bin/node ... gateway --port 18789)
• Gateway bind: loopback 127.0.0.1:18789

Symptoms

• openclaw cron status / openclaw cron list often fail:
  • Error: gateway timeout after 10000ms
• With larger timeout they succeed but are extremely slow:
  • openclaw cron list --timeout 120000 succeeded but took ~114s (observed)

Evidence from gateway logs

From /tmp/openclaw/openclaw-2026-02-04.log, there are many successful but very slow responses:

• ⇄ res ✓ cron.list 212603ms ... (212s)
• ⇄ res ✓ cron.list 204194ms ...
• ⇄ res ✓ cron.status 118209ms ...

Example (same window also shows long main-lane work):

• ⇄ res ✓ cron.list 212603ms ... (line ~7596)
• lane task done: lane=main durationMs=511198 active=0 queued=0 (line ~7612)

Additionally, one one-shot job state repeatedly showed:

• lastStatus=skipped
• lastError="timeout waiting for main lane to become idle"

This suggests cron RPC handlers and/or at-job execution may be blocked behind (or coordinated with) main lane idleness, causing cron RPC starvation when the agent is busy.

Expected

• cron.status / cron.list should be fast (sub-second to a few seconds) and not depend on the main agent lane.

Questions

1. Are cron.status / cron.list handled on the main lane today?
2. Is there known lane/lock contention between the cron subsystem and the session/agent lane?
3. Are there mitigations (config knobs / lane assignment / caching cron snapshot) to avoid starving cron RPC?

Workarounds currently used

• Use --timeout 120000 for cron CLI commands.
• For health checks, read ~/.openclaw/cron/jobs.json directly when RPC wedges.