PingingPing
Nightly backup
nightly-backup
Pinged6 h ago
checked 1 min ago
Heartbeat monitoring
A minted URL for each job that runs on its own schedule, and an alert when a run does not call it in time.
Included on every plan.
PlansFreeProBusiness
| What breaks | hostnag |
|---|---|
| A scheduled job stops runningcron, backups, workers: a heartbeat that goes quiet | Yes |
On the board
Sample plates of this type, drawn by the board itself.Acme Retail
03Sample dataMissedPing
Invoice sync
invoice-sync
Missed32 min
checked 1 min ago
UpReport
Warehouse scanner
warehouse-scanner
Reported3 min ago
checked 1 min ago
How it works
- 01
Add the job
A cron run, a backup, a worker: hostnag mints a ping URL for it, no installed software involved.
- 02
Say how often it should call
An expected interval and a grace period before a missed ping counts as overdue.
- 03
The job calls the URL when it finishes
One line at the end of the job's own script. Only the job can say it finished, so it pings for itself; a Beacon feeds links, status reports and certificate reports, never a job's heartbeat.
- 04
Silence is the alert
No ping past the deadline raises down, and hostnag says exactly how many minutes overdue it is.
What you'll be told
What you get told
- Ping
[Critical] Nightly backup missed its heartbeat
nightly-backup has not pinged in time: 25 min overdue. The job may have stopped or hung, or it cannot reach hostnag.
- Ping
[Recovered] Nightly backup is pinging again after 1 h 15 min
nightly-backup pinged again. It was silent for 1 h 15 min.
- Cert
[Warning] Checks failing for api.example.com
We could not check api.example.com on three consecutive attempts: connect_failed: ECONNREFUSED 203.0.113.9:443.
Sample alerts, rendered by the same code that writes your email and Slack messages.
Settings
- Target
- a job name, turned into a slug
- Expected interval
- any number of minutes
- Grace period
- 0 to 1440min
- Status reports
- up, down or degraded with a message, optional cadence
- Fed by
- the ping URL directly from the job
Questions
Do I need to install anything on the job's host?
No. One line that calls a URL, or the Beacon, a plain script one line installs under your own scheduler, which you can read and remove yourself.
What if the job wants to report degraded instead of missing entirely?
A status report can say up, down or degraded with its own message; silence past its cadence is a "could not check", never a down.
Related
Link monitoringAn office or ISP connection watched from both ends, so a filtered port and a dead line never read the same.Certificate monitoringEvery certificate hostnag can reach is watched for expiry, a swapped chain and a renewal that should already have happened; the ones it cannot reach get a URL to report to instead.
Read
How to monitor cron jobs with heartbeats: one curl line and a grace periodA cron job has no address to check, so it checks in instead. One curl -fsS line at the end of the job, and a grace period sized to its slowest run.How to monitor an office internet link from both ends, so a filtered port and a dead line read differentlyReach the gateway from the internet and have a device inside check in, so a filtered port, a dead line and a new public address each read differently.
Add a host. Hear about it first.
Start free