Healthchecks — Cron Job Monitoring with Smart Alerts

What it is

Healthchecks is a self-hosted monitoring service for cron jobs and scheduled tasks. Instead of checking whether a service is up, it checks whether your periodic jobs are running on schedule. When a job misses its expected ping, Healthchecks sends an alert.

It serves system administrators, DevOps engineers, and backend developers who run scheduled tasks (backups, data syncs, report generation) and need to know immediately when something stops working silently.

How it saves time or tokens

Failed cron jobs are silent failures. Without monitoring, you might not discover a broken backup job until you need a restore. Healthchecks converts these invisible failures into immediate alerts via email, Slack, PagerDuty, or webhooks. The ping-based design requires only a single HTTP request or curl call appended to your existing scripts.

How to use

1. Create a check in Healthchecks with the expected schedule (cron expression or simple interval).
2. Copy the unique ping URL and add a curl call to the end of your cron job script.
3. Configure notification channels (email, Slack, PagerDuty) to receive alerts when pings are missed.

Example

# Add a ping to your existing cron job
# crontab -e
0 2 * * * /usr/local/bin/backup.sh && curl -fsS -m 10 --retry 5 https://hc-ping.com/your-uuid-here

# Python script with Healthchecks ping
import requests
import subprocess

try:
    subprocess.run(['python', 'etl_pipeline.py'], check=True)
    requests.get('https://hc-ping.com/your-uuid-here')
except Exception as e:
    requests.get('https://hc-ping.com/your-uuid-here/fail')
    raise

Related on TokRepo

• AI Tools for Monitoring — Monitoring and observability tools for infrastructure
• AI Tools for DevOps — DevOps automation and infrastructure management

Common pitfalls

• Setting the grace period too short, causing false alerts when jobs take slightly longer than expected.
• Pinging at the start of the job instead of the end, which hides jobs that start but fail to complete.
• Forgetting to handle the failure ping endpoint for jobs that should report errors explicitly.