Cron Management

cron-management — OpenClaw Cron Job Manager

Structured commands for managing cron jobs without raw JSON patches.

Commands

Maintenance Mode

Emergency stop that works even if OpenClaw is down. Jobs check for /tmp/cron-paused at startup:

if [ -f /tmp/cron-paused ]; then
  echo "Cron paused at $(cat /tmp/cron-paused)" >> /tmp/cron-skipped.log
  exit 0  # or reply NO_REPLY for agentTurn jobs
fi

Toggle: cronctl maintenance on|off

Script

• scripts/cronctl.sh — Main CLI wrapper around openclaw cron

Integration with Job Payloads

To add maintenance mode checks to existing jobs, add this at the top of each cron payload:

For shell-based payloads:

if [ -f /tmp/cron-paused ]; then
  echo "$(date -Iseconds) — $JOB_NAME skipped (maintenance mode)" >> /tmp/cron-skipped.log
  exit 0
fi

For agentTurn payloads (OpenClaw): Add to the beginning of the message text:

If /tmp/cron-paused exists, reply NO_REPLY and exit.

For Python scripts:

import os
import sys
from datetime import datetime

PAUSE_FILE = "/tmp/cron-paused"
SKIP_LOG = "/tmp/cron-skipped.log"

def check_maintenance_mode():
    if os.path.exists(PAUSE_FILE):
        with open(SKIP_LOG, "a") as f:
            f.write(f"{datetime.now().isoformat()} — {sys.argv[0]} skipped (maintenance mode)\n")
        sys.exit(0)

check_maintenance_mode()

For Node.js scripts:

const fs = require('fs');
const path = require('path');

const PAUSE_FILE = '/tmp/cron-paused';
const SKIP_LOG = '/tmp/cron-skipped.log';

function checkMaintenanceMode() {
  if (fs.existsSync(PAUSE_FILE)) {
    const entry = `${new Date().toISOString()} — ${path.basename(process.argv[1])} skipped (maintenance mode)\n`;
    fs.appendFileSync(SKIP_LOG, entry);
    process.exit(0);
  }
}

checkMaintenanceMode();

Rollout Strategy

1. New jobs: Include maintenance check from the start
2. Existing jobs: Update incrementally, starting with most frequent (every 30 min, hourly)
3. Verification: Enable maintenance mode, wait for next trigger, check /tmp/cron-skipped.log

Health Dashboard

cronctl status shows:

• Total jobs, enabled/disabled counts
• Jobs that failed in last 24h (consecutiveErrors > 0)
• Jobs disabled but not in maintenance mode (unexpected state)
• Jobs whose last run was >2x their interval (overdue)

Examples

# Quick status overview
cronctl status

# Pause just the blog writer for debugging
cronctl pause cloudy-blog-writer

# Emergency stop everything
cronctl maintenance on

# Resume everything after fix
cronctl maintenance off
cronctl resume-all

# Check why a job keeps failing
cronctl health graph-memory-queue-sync

Implementation Details

See memory-bank/implementation-details/cron-management.md for:

• Architecture (two-layer design)
• Rollout strategy
• Error handling
• State files and security
• Future enhancements