hostnag

Uptime monitoring

Websites, APIs and anything that answers on a port, checked every minute from up to five places, with a failure confirmed before anyone is paged.

Start freeSee pricing

Included on every plan.

PlansFreeProBusiness
What Uptime monitoring catches
What breakshostnag
A host goes downYes

On the board

The location board and the day’s recorder from a monitor page. Hover a cell.

Locations

1 of 5 failed
US Eastiad1
84msHTTP 200
US Westsfo1
131msHTTP 200
Frankfurtfra1
unexpected_status: 503
2 attempts
Tokyohnd1
212msHTTP 200
Sydneysyd1
246msHTTP 200
Last 24 hours, UTC
  • Up
  • Blip
  • Down
  • No runs
012502500p95, ms
00:0006:0012:0018:00

How it works

  1. 01

    Add the address

    A URL for HTTP, or a host and port for TCP: a VPN gateway, a mail server, a database, a game server. Private and loopback addresses are refused; hostnag watches from the cloud. An IP or host alone can be pinged.

  2. 02

    Say what a good answer looks like

    A status code range, a keyword that must be present or absent (case-insensitive unless you say otherwise), a regex, up to ten JSON assertions on the body, a redirect policy and a latency threshold.

  3. 03

    Pick where and how often

    Every 5 minutes from one location on Free; every 1 minute from up to 3 (Pro) or 5 (Business) of US East, US West, Frankfurt, Tokyo, Sydney.

  4. 04

    Down is confirmed, then you are told

    One failed run can be a blip, so hostnag needs 2 confirmations, a second location or a second consecutive run, before it opens an episode and nags your channels. A recovery notice follows when the host is back.

What you'll be told

What you get told

  1. [Critical] https://www.example.com/ is down

    Down from 2 of 3 locations (US East, Frankfurt). Last error: unexpected_status: 503.

    HTTP

  2. [Warning] https://www.example.com/ is responding slowly (2400 ms)

    https://www.example.com/ is up but slow: 2400 ms median across locations, above the threshold you set.

    HTTP

  3. [Warning] https://www.example.com/ p95 is 1840 ms over 24 h (SLO 800 ms)

    https://www.example.com/ answered within 1840 ms for 95% of 96 checks in the last 24 h, above the 800 ms you set. It is still up.

    HTTP

  4. [Recovered] https://www.example.com/ is back up after 1 h 15 min

    https://www.example.com/ answered again from every location. It was down for 1 h 15 min.

    HTTP

Sample alerts, rendered by the same code that writes your email and Slack messages.

Settings

Expected status
a code or a range, 200 to 299 by default
Keyword
present or absent, case-insensitive by default, regex optional
JSON assertions
up to 10 path rules: equals, contains, exists, is absent, greater than, less than
Method and body
GET or POST with a body and custom headers (headers are stored in plaintext; the form says so)
Redirects
follow, or fail on any redirect
Latency
a per-run threshold that marks the monitor degraded, and a daily p95 SLO
Timeout
15s
Interval
5 min on Free, 1 min on Pro and Business

Questions

Do you ping with ICMP?

Yes. A Ping monitor sends ICMP echo requests from probes of our own in US East, US West, Frankfurt, Tokyo, Sydney, several packets a run, and records replies, loss and round-trip time per place. A Beacon can ping from inside your network too.

Will I get paged when one location has a bad minute?

No. A failure needs 2 confirmations, from a second location or a second consecutive run, before an episode opens.

Can I watch a private host?

Yes, with a Beacon. One line installs a plain script on a box in that network, the check runs from there, and hostnag judges what it saw. A host can also report its own status through a status report URL.

Related

AlertingEach alert goes to the channels its severity calls for, waits out quiet hours and maintenance, and escalates until someone acknowledges it.Heartbeat monitoringA minted URL for each job that runs on its own schedule, and an alert when a run does not call it in time.Vendor status monitoringYour vendors' incidents on the same board as your own, read from the status pages they already publish.

Read

How to monitor a private host with a Beacon: one readable script, run by your own schedulerWatch an intranet host or an internal certificate from inside your network with one plain script, scheduled by cron, launchd or Task Scheduler.Ping is not uptime monitoring: what an ICMP reply proves, and why a failure needs a second witnessAn ICMP reply proves a machine is there. A TCP connect on the right port is the first real test, and a failure needs a second place to confirm it.

Add a host. Hear about it first.

Start free