CRON JOB MONITORING

See every job. Catch every failure.

Cron jobs rarely fail loudly. They may exit 0 with an empty result, start ten minutes late, or quietly stop running.

  • Get an alert when a job fails, starts late, or never starts.
  • See the status, metrics, and logs for every run.
  • Monitor cron jobs, background workers, and scheduled agent tasks.
cronitor.io/llms.txt — agent quick start
scheduled jobs · 6 monitored · 43 executions 02:00 → 08:00 · now 07:04
JOB
NOW
CONCURRENCY peak 6 · 15m buckets
6 JOBS FIRE AT 03:00
Disney
Johnson & Johnson
Reddit
Shutterstock
Genentech
Yelp
GOV.UK
Razer
Rakuten
AB InBev
Chime
Monzo
monday.com
GoodRx
Zendesk
Meesho
stripe-payout-sync · */5 * * * * median 41.7s · 95.5% success
DURATION PER EXECUTION44 SCHEDULED · 3h40m WINDOW
▲ALERT THRESHOLD 60.0s
04:0007:40
ALERTS
—watching 44 executions
Slack#ops-alerts
1 missed · 1 over threshold · 1 failed
ALERTS

Get alerted when a job fails, starts late, or runs too long

A job that crashed, one that never started, and one that's been hanging for an hour all look the same from outside: silence. Cronitor tells you whether the run failed, never started, or is taking too long, so you know where to investigate.

  • Get alerted when an expected run doesn't start.
  • Let one flaky run slide and get paged when it keeps failing.
LOGS · METRICS · RETENTION

Read any run's output without logging into the server

Every run keeps what it printed: the error, the last lines of output, how long it took. Compare it with a run that worked, even months later.

  • Open any run, failed or slow, and read exactly what it printed.
  • Spot the job that gets slower every week before it starts timing out.
  • Compare any run with the last one that worked, up to 12 months back.
send-invoices · 0 * * * * 8,760 executions · 12 mo retained
SUCCESS RATE %
100.0
95.0
DURATION p50 / p95 · SECONDS
110
▲THRESHOLD 60.0s
CUSTOM METRIC invoices_sent
p50 PER RUN · 160
ALERTS FIRED
▲p95 2.2× BASELINE
CURSOR
SEP2025
SUCCESS99.9%
p5040.6s
p9544.2s
invoices_sent104
ALERTS0
SLOWEST RUN ●48.9s#335
STDOUT + STDERRrun #33514 SEP 2025 · 22:00 UTC LAST 4 OF 12 LINES
READING 12 MONTHS sep 2025 → aug 2026

Your jobs, your stack, a few lines of code

Whatever runs your background work, there's an open-source SDK or integration for it. Add it, and every run reports its status, duration, and output. Using something we don't cover? Send events to the HTTP API with one request.

Start for free
crontabkuberneteswindowsdockerpythoncelerynoderubysidekiqphplaravelhttp api
# Install CronitorCLI on the box that runs your jobs
$ curl https://cronitor.io/install-linux?sudo=1 -H "API-KEY: <API-KEY>" | sh

# Walk the crontab and add monitoring to the jobs you choose
$ cronitor sync
----► Scanning for cron jobs...
 |--► Found 15 jobs in user "ec2-user" crontab
 |--► send-invoices        │ ✔ Monitor this job
----► Sync complete ✔

# Your crontab comes back instrumented. Status, duration and output
# are reported for you — the job itself never changes.
0 * * * * CRONITOR=send-invoices /usr/local/bin/send-invoices.sh
HOW IT WORKS* * * * *

See what ran, catch what didn't, and alert the right person

Cronitor records what each job did, compares it with the schedule, and alerts whoever is on call when the two do not match.

01INSTRUMENTwhat ran
Know what's scheduled, everywhere

Cron on one box, Kubernetes CronJobs on another, Sidekiq inside the app: see them all in one list, with when each last ran and when it's due next.

Logs land on the run that wrote them

Cronitor captures stdout and stderr as the job emits them and stores the output against that execution, so the output is there when you open the run.

OUTa timestamped record of every execution
→
02DETECTwhat didn't
The run that never started

A job that never starts can't report its own failure. Cronitor knows the schedule, so when a run doesn't show up on time, you get the alert.

The run that took too long

A job that usually takes two minutes and is still going at forty is a problem. Cronitor compares each run with that job's own history and alerts you when it's running long.

OUTa named failure, tied to one execution
→
03RESOLVEwho to tell
Alert rules you can live with

Send each job's alerts to the Slack channel or PagerDuty rotation that owns it, and decide how many failures in a row it takes to page.

12 months of evidence

Success rate, duration percentiles and custom metrics, retained a full year on dashboards that load fast enough to actually use during an incident.

OUTthe right person, with the evidence
FROM THE FIELD* * * * *
FAQ
01What is cron job monitoring?

Cron job monitoring verifies that your scheduled jobs start on time, finish successfully, and complete within a normal duration. Cronitor records every execution with its status, duration, metrics, and captured output, and alerts you the moment a job fails or an expected run never starts.

02How does Cronitor know when a cron job fails?

Each job reports its lifecycle to Cronitor, typically by wrapping the scheduled command with cronitor exec or by sending simple telemetry pings from your code. If a run reports a failure, exits with a non-zero code, or the expected run never arrives, Cronitor triggers an alert to email, Slack, PagerDuty, or any other channel you configure.

03Do I need to change my code to monitor a cron job?

No code changes are required. Wrapping the command in your crontab with cronitor exec is a one-line change per job and captures start, completion, exit status, and output automatically. If you prefer, you can instrument jobs directly with an HTTP ping or one of Cronitor's SDKs instead.

04How do I monitor an AI agent launched by cron?

Create a Job monitor with the agent tag, then wrap the scheduled command with cronitor exec MONITOR_KEY -- COMMAND. Cronitor records when Codex, Claude Code, or another agent starts, completes, or fails and alerts you when the cron-driven run is missing.

05What does Cronitor record for a bounded agent job?

Every invocation has a run, complete, or fail outcome plus its duration and captured output. Telemetry can also carry custom metrics, such as tokens or cost, as metric=name:value parameters. Cronitor charts them and alerts you when a value breaks an assertion.

* * * * *

Monitoring for Developers