Uptime monitoring · incident alerting

The klaxon sounds before your customers do.

Klaxon checks your endpoints from twelve regions around the clock and pages whoever is on call the moment a failure is confirmed. Slack, SMS, and a phone call that keeps ringing until someone answers. Median time from failed check to first page: eight seconds.

4,182,554
Checks run today
23
Incidents caught today
8.4s
Median time-to-alert

Alert wire

12:00:00 UTC

  1. Started
    11:56 UTC
    Region
    FRA · Frankfurt
    Response
    HTTP 503
    Trace
    212 · 305 · 1.9k · — ms
    Ack
    M. Okafor · 42 s
  2. Started
    11:51 UTC
    Region
    LHR · London
    Response
    HTTP 200
    Trace
    191 · 187 · 204 · 183 ms
    Ack
    auto-resolved
  3. Started
    11:49 UTC
    Region
    IAD · N. Virginia
    Response
    HTTP 200
    Trace
    780 · 912 · 1.1k · 1.2k ms
    Ack
    J. Lindqvist · 2 m
  4. Started
    11:47 UTC
    Region
    AMS · Amsterdam
    Response
    TLS 1.3 · SNI ok
    Trace
    118 · 122 · 120 · 119 ms
    Ack
    auto-resolved
  5. Started
    11:42 UTC
    Region
    SYD · Sydney
    Response
    heartbeat POST
    Trace
    44 · 51 · 39 · 47 ms
    Ack
    auto-resolved
  6. Started
    11:36 UTC
    Region
    GRU · São Paulo
    Response
    timeout
    Trace
    240 · 310 · — · — ms
    Ack
    R. Patel · 27 s
  7. Started
    11:20 UTC
    Region
    SIN · Singapore
    Response
    HTTP 200
    Trace
    505 · 462 · 430 · 421 ms
    Ack
    auto-resolved

12 regions · checks every 30 s · all times UTC

Incidents now

Live · 22m checkout-eu Elevated 5xx from FRA and LHR HTTP 503
Live · 9m webhooks-eu Delivery latency above threshold 4.1 s
Resolved 11:55 api.acme.dev Recovered, all regions green 3m 41s
Resolved 11:12 search p95 back under 800 ms 12m
Resolved 10:48 cron: nightly-backup Heartbeat received 6m late
Resolved 10:03 auth.acme.dev Certificate renewed 90 days
Resolved 09:41 cdn-edge Recovered after failover 4m 02s

Thirty-day view

The uptime board.

Every service, last 30 days. The hotter the cell, the rougher its month. Hover or tab to any cell for the uptime figure; the incident counts stay on the board.

The board redraws on every check. On-call accounts can pin any cell to a public status page.

Signal review

Noisiest services, last 30 days.

Where the alerts came from. On-call accounts get the full ledger for every service on the wire: every check, every page, who acknowledged and how fast.

  1. 01 checkout-eu 9 incidents · 3h 12m down 34 alerts
  2. 02 webhooks-eu 7 incidents · 1h 48m down 28 alerts
  3. 03 search 5 incidents · 54m down 19 alerts
  4. 04 payments-api 4 incidents · 31m down 14 alerts
  5. 05 cdn-edge 3 incidents · 12m down 9 alerts
  6. 06 mobile-api 2 incidents · 9m down 7 alerts

Standard procedure

Three moves between you and a quiet night.

  1. Monitor

    Paste a URL and pick your regions. HTTP, TCP, ping and cron heartbeats, checked as often as every 30 seconds from twelve cities. A failed check is re-run from other regions before anything makes a sound, so a wobbly route never wakes anyone.

  2. Escalate

    Set thresholds once, and the escalation path with them: who is on call, in what order, and how long before it climbs the chain. Rotations live in the same place, so handover happens on schedule without anyone passing a pager across a desk.

  3. Get paged

    Slack, SMS, push, and a phone call that keeps ringing until someone acknowledges. Median time from failed check to first page is eight seconds, wherever in the world you happen to be asleep.

Access

Two ways to stand the watch.

Level 1 · Watch

£0free forever, no card

  • 10 monitors, checked every 3 minutes
  • Checks from 3 regions
  • Email and Slack alerts
  • One hosted status page
  • 90 days of history
Start monitoring

Gets loud

Level 2 · On-call

£15per month, cancel any time

  • Everything in Watch
  • Unlimited monitors, 30-second checks
  • All twelve regions
  • Phone and SMS paging
  • On-call rotations and escalation chains
  • Status pages on your own domain
  • API, Terraform and CSV export
  • Priority support desk
Go on call

Prices exclude VAT. Unlimited teammates on both levels: you pay for the monitoring, never for seats.

Questions

From the night shift.

How often do the checks run?

Every 30 seconds on On-call, every 3 minutes on Watch, from every region you enable. Any monitor can run a lighter schedule if you prefer; the pager only ever cares about confirmed failures.

Where do the checks run from?

Twelve cities across five continents, London to Singapore to São Paulo. Pick any subset per monitor: a service for European customers should be checked from Europe, and a global one from everywhere its customers are.

How do you handle false positives?

A failed check never pages on its own. It is re-run immediately from two other regions, and only a confirmed failure sounds the klaxon. Blips that recover inside the confirmation window are logged on the incident timeline, not sent to your phone at three in the morning.

Do you host status pages?

Yes, on both levels. Watch includes one hosted page; On-call adds unlimited pages on your own domain, fed straight from the same checks, with incident notes you publish from the app while you fix things.

How do on-call rotations work?

Build a schedule, set the order, and Klaxon pages whoever is on shift. Acknowledge from the app, by SMS reply, or by pressing 4 on the call itself. An unacknowledged alert climbs the escalation chain until a human answers; nothing is ever quietly dropped.