ETL Job Monitoring

Your ETL job completed.
It exported 0 rows.

Data pipelines fail in ways that look like success. Exit 0, ping sent, dashboard green — while your downstream tables sit empty. DeadManCheck catches it.

Start monitoring free → See quickstart
Standard monitoring
✓ ETL ran at 04:00
✓ Exit code 0
✓ Ping received
Rows inserted: 0  ·  upstream API token expired
Alert sent: no
DeadManCheck
✓ ETL ran at 04:00
✓ Exit code 0
✓ Ping received
rows_inserted: 0  ·  assertion: rows_inserted > 0
⚠ Assertion failed — alert sent

How ETL jobs fail silently

⚠
API token expired — upstream returned empty response
Job handled the empty response gracefully. Exited 0. Inserted nothing. Your tables are now a day behind.
⚠
Schema change upstream broke parsing — 0 records matched
The source added a field. Your parser skipped every row. Exit 0. You won't notice until someone queries stale data.
⌛
10× more data than usual — job took 3 hours instead of 10 minutes
The pipeline succeeded. But it's now competing with your morning workload and your users are hitting stale reads.
⚠
Deduplication logic removed all rows — net inserts: 0
A logic change meant every incoming record was treated as a duplicate. The job ran fine. Nothing landed in your warehouse.

Monitor your ETL in 3 lines

Send row counts with your ping. Set assertions. Get alerted the moment your pipeline produces nothing.

# Python
import requests
requests.get("https://deadmancheck.io/ping/your-token/start")
result = run_etl_pipeline()
requests.post("https://deadmancheck.io/ping/your-token/end",
json={
"rows_inserted": result.inserted,
"rows_failed": result.failed,
"source": result.source_name
})
# Bash
curl -s https://deadmancheck.io/ping/your-token/start
ROWS=$(python etl.py --count-only)
curl -s -X POST https://deadmancheck.io/ping/your-token/end \
-d "{\"rows_inserted\": $ROWS}"
rows_inserted > 0
rows_failed == 0
source != "empty"

Set these assertions in the dashboard. If any fail, you get alerted immediately.

Everything you need for ETL monitoring

✓
Missing pipeline alerts
Know immediately if your ETL doesn't run on schedule — before downstream queries hit stale data.
✓
Duration monitoring
Automatic baseline learning. Alert when your pipeline takes 10× longer than normal — before it impacts your users.
✓
Did it actually insert any data?
Validate row counts, error counts, source names — anything your ETL produces. Alert the moment your pipeline returns empty.
✓
Works with any pipeline stack
Airflow, dbt, Prefect, Luigi, custom Python/Bash scripts — anything that can make an HTTP request.

Related guides

Feature
Cron job output monitoring
How output assertions work
Feature
Monitor long-running cron jobs
Catch pipelines that take far too long
Use case
Backup monitoring
Monitor backup jobs for silent failures

Stop trusting "exit 0" on your data pipeline

Free plan. 5 monitors. No card required. Set up in 5 minutes.

Start monitoring free →

Or self-host for free on GitHub