Docs
Pick your platform for a step-by-step guide, or read the reference below. Every check's Setup tab also has copy-paste snippets with its own URL filled in.
Guides
- GitHub ActionsGitHub Actions scheduled workflow not runningThe 60-day rule, delayed runs, and the Lateping action that catches both.
- Cloudflare WorkersMonitor Cloudflare Workers Cron TriggersPing from scheduled(), and report failures from the catch.
- crontabFind out when a crontab job fails silentlyOne curl after the command, plus cron's PATH and MAILTO.
- Vercel CronMonitor Vercel Cron JobsPing from the route the cron calls, and report failures.
- pg_cronMonitor pg_cron jobs in PostgresPing with pg_net after the job commits, and on failure.
- Kubernetes CronJobsMonitor Kubernetes CronJobsWrap the command with pings, and know why runs get skipped.
- systemd timersMonitor systemd timersA small ping script, run before and after the service.
- Backup jobsGet alerted when a backup job stops runningFail loudly on any step, check the dump isn't empty, then ping.
- Laravel schedulerMonitor the Laravel schedulerpingOnSuccess, pingOnFailure, and a heartbeat for schedule:run.
- Dead man's switchA dead man's switch for cron jobsWhy silence should be the alarm, and how to set one up.
- Did my cron job fail?How to know if a cron job failedExit codes, cron logs, MAILTO, and the one failure none of them catch.
- anacronCron job didn't run because the computer was off (anacron)Laptops and small servers skip cron jobs while off. Use anacron, then monitor it.
- Cron exit codesHow to get an alert with the exit code when a cron job failsSend a cron job's exit code to a ping URL and see which code failed in the alert.
- overlapping runsCron job running twice at the same time (overlapping runs)Stop a slow cron job from starting twice with flock, and see when it was skipped.
- cron job timeoutCron job hangs and never finishes (kill it with timeout)Stop a hung cron job with timeout, and get an alert when it was killed.
- cron job timezoneCron job runs at the wrong time (timezone and daylight saving)Check cron's timezone, set CRON_TZ, and avoid the daylight saving gap.
- cron job retryHow to retry a failed cron jobA retry loop for cron, with one ping at the end instead of one per attempt.
Ping URLs
Each check has its own URL. Your job calls it; nothing else is needed. Any HTTP method works, usually GET or POST.
| URL | Records | Alerts |
|---|---|---|
/p/<check-id> | A successful run. | If the check was late or failed, we tell you it recovered. |
/p/<check-id>/start | That a run began. | None. |
/p/<check-id>/fail | A failed run. | Within a minute, without waiting for the grace period. |
/p/<check-id>/<exit code> | The job's exit code, 0 to 255. 0 is a success; anything else is a failed run, and the code is shown on the alert. | As for /fail when the code is not 0; as for a success when it is 0. |
Run this after your job finishes:
curl -fsS -m 10 --retry 5 -o /dev/null https://lateping.com/p/<check-id>We don't store request bodies or IP addresses. Each ping keeps its time, its kind and a country code.
Responses
200 OK: the ping was recorded.404 Not found: no check has that ID, or the URL is malformed. Withcurl -f, a wrong ID fails the command instead of passing silently.500 Error: our side failed.--retry 5tries again; a ping that never lands shows up as a late check, not a missing alert.
Timing a run
Call /start before the job and the plain URL after it. Lateping records both, so you can see when each run began and finished. On Pro and above, the check's Analytics tab shows run times, averages and charts, and exports them as CSV.
curl -fsS -m 10 --retry 5 -o /dev/null https://lateping.com/p/<check-id>/start
curl -fsS -m 10 --retry 5 -o /dev/null https://lateping.com/p/<check-id>Put your job's own commands between the two lines. To report a failure straight away, call/fail when the job exits non-zero; the guides show how on each platform.
Schedules and grace
- Every N minutes: the next ping is due N minutes after the last one.
- Cron expression with a timezone: the next ping is due at the next scheduled time, in that timezone. GitHub Actions and Cloudflare Cron Triggers run in UTC.
- Grace: how long we wait after a ping is due before alerting. Allow for the job's run time and any queue delay. 30 minutes suits most nightly jobs.
A new check starts with its first ping. Until then, it never alerts.
Alerts
- Late: no ping arrived before the grace period ended.
- Failed: your job called
/fail. - Recovered: a successful ping arrived after a late or failed alert.
Alerts go to your account email. From the panel's Destinations tab you can add more email addresses on any plan, and Slack, Discord or a signed webhook on Pro and above.
Pausing a check
Pause a check from its Settings tab, for example during maintenance. A paused check doesn't alert when pings stop. The next successful ping starts it again, and a /fail ping still alerts.