Move EasyCron, cron-job.org, and GitHub Actions timers onto a scheduler you host

EasyCron, cron-job.org, and GitHub Actions scheduled workflows all sit outside the box that actually does the work. That path holds URLs, bearer tokens, and run history. When the bill is monthly, you are renting a timer. NexusCron is the same architecture on your host: schedule, HTTP fire, retry, SQLite logs.

Docker Compose

Put the data volume on the same disk you already back up. Secrets stay in cron.env, not in the image.

services:
  nexuscron:
    image: nexuscron:latest
    restart: unless-stopped
    ports:
      - "127.0.0.1:8787:8787"
    environment:
      TZ: Europe/Madrid
      NEXUSCRON_DATA: /data
      NEXUS_TELEMETRY: "off"
      CRON_ADMIN_TOKEN: \${CRON_ADMIN_TOKEN}
    volumes:
      - ./data:/data
      - ./jobs.json:/app/jobs.json:ro
      - ./cron.env:/run/secrets/env:ro

Production environment

Set TZ to the zone your cron expressions assume. Rotate CRON_ADMIN_TOKEN like any other admin secret.

# cron.env (chmod 600)
CRON_ADMIN_TOKEN=replace-me
TZ=Europe/Madrid
# Optional: HMAC on inbound health pings
CRON_PING_SECRET=replace-me
# Optional: default timeout for HTTP jobs
CRON_HTTP_TIMEOUT_MS=15000

jobs.json

One object per hosted timer. Keep the same cron string and URL so cutover is a pause on the vendor side, not an app change.

{
  "jobs": [
    {
      "id": "billing-reconcile",
      "cron": "15 3 * * *",
      "timezone": "Europe/Madrid",
      "http": {
        "url": "http://billing.internal/jobs/reconcile",
        "method": "POST",
        "headers": { "Authorization": "Bearer ${BILLING_TOKEN}" }
      },
      "retry": { "max": 3, "backoff_ms": 2000 }
    },
    {
      "id": "uptime-ping",
      "cron": "*/5 * * * *",
      "http": {
        "url": "https://app.yourdomain.com/internal/health",
        "method": "GET"
      }
    }
  ]
}

Cutover without missed pings

1. Export every hosted job (expression, URL, headers).
2. Bring NexusCron up. Confirm /health.
3. Load jobs.json. Run one job immediately.
4. Dual-run for one interval if the endpoint is idempotent.
5. Pause EasyCron / cron-job.org. Keep the CSV until you trust SQLite.

# 1. Health
curl -sS -o /dev/null -w "%{http_code}\\n" http://127.0.0.1:8787/health

# 2. Load jobs and fire one now
curl -sS -X POST http://127.0.0.1:8787/api/jobs/uptime-ping/run \\
  -H "Authorization: Bearer $CRON_ADMIN_TOKEN"

# 3. After a clean day, pause EasyCron / cron-job.org entries
# Keep their export CSV next to jobs.json until you trust the table

Related setup page: Self-hosted cron alternative.