Copy, paste, monitored.
Every guide below ends in a working snippet — a crontab line, a manifest, a unit file, a try/except — against the real ping API. Pick your environment, paste the block, and the next silent failure has a pager.
The ping API, in six lines
Every monitor has one URL; suffixes change the meaning. GET, HEAD, or
POST — a POSTed body (up to 10 KB) is stored with the ping, and
?rid= pairs a start with its finish to record run
duration, and ties a run's progress steps to it.
/step is the only suffix that reports progress rather than a
result. It says "this run is still moving": it resets the monitor's stall
clock and records the step on the run timeline, but it does not change the
monitor's status, does not count as a check-in for silence detection, and
never extends the run budget. It alerts only if you opt in by setting
step_timeout_s on the monitor — leave it unset and steps are
simply recorded. See
hung-job detection for the full pattern.
Pick your environment
Heartbeats for things that run, probes for things that serve, and the alerting reference that shapes what lands in your channel.
Cron jobs
The flagship one-liner: append a curl to the crontab line; failures, hangs, and vanished schedules all alert.
job && curl -fsS -m 10 --retry 3 …/<id> Heartbeat · KubernetesKubernetes CronJobs
A copy-paste manifest — success and failure pings from the job command, no operator, nothing installed in the cluster.
if /app/job.sh; then curl "$LASTPING_URL"; else … Heartbeat · systemdsystemd timers
ExecStartPost pings success, OnFailure pings failure, absence covers the timer that stopped firing. Three unit files.
ExecStartPost=/usr/bin/curl -fsS …/<id> Heartbeat · patternBackups
Start → work → success/fail, with the error log POSTed on failure and run duration recorded. The anti-horror-story.
curl …/start · pg_dump … · curl …/fail Heartbeat · in codePython
A try/except with requests or plain stdlib — ping on success, POST the exception on failure, re-raise for the exit code.
requests.post(f"{PING}/fail", data=str(exc)) Heartbeat · in codeNode.js
Built-in fetch, zero dependencies — ESM and CommonJS variants, with the stack trace attached to the fail ping.
await fetch(`${PING}/fail`, { method: "POST" … }) CI/CD · webhookGitHub Actions
One signed webhook, zero YAML changes — catches failed, hung, and never-started runs, naming the step that broke.
Settings → Webhooks → workflow_run + workflow_job CI/CD · webhookGitLab CI
One Pipeline-events webhook, zero .gitlab-ci.yml changes — catches failed, pending-stuck, and never-started pipelines with failure detail.
Settings → Webhooks → Pipeline events CI/CD · webhookJenkins
One Notification-plugin endpoint, no Jenkinsfile changes — catches failed, queue-stuck, and never-started builds, even a dead controller.
Job Notifications → JSON / HTTP Probe · uptimeHTTP & uptime
LastPing polls your URL every 30 s–24 h: status code, latency, optional body keyword — with the exact reason on the incident.
status 500, expected 200 · body match failed Reference · alertingAlert events & templates
down, recovery, fail, every-run, success, started, blocked, note — the routing matrix and per-event message templates with 20 variables.
[LastPing] {check_name} {event} ({cause})Cloud & serverless
CI/CD platforms
AI agents
Deciding between tools first? Read the honest comparisons with Healthchecks.io and Cronitor, see how LastPing monitors itself, or go straight to the full documentation.
Somewhere in your stack, a job is failing politely.
Every snippet on these pages is wired in minutes — and LastPing monitors itself the same way.