Skip to main content

Features

Everything between
a signal and a fix.

Checks, regions, incidents, maintenance windows, certificates, and routing — designed so the alert that reaches you is one you actually needed to see.

Example GeeksUptime console: five monitors across eight regions. Checkout is down in two regions with an open incident routed to Slack, email, SMS, and a webhook.

geeksuptime.com/app · workspace: production 8 regions
Up
47
Down
1
Unknown
1
Availability 30d
99.97%
  • Marketing sitehttps://acme.io
    182 ms
    UP
  • Public APIhttps://api.acme.io/health
    96 ms
    UP
  • Checkouthttps://checkout.acme.io
    timeout
    Down 4m
  • Auth serviceping · auth.internal
    12 ms
    UP
  • Media CDNhttps://cdn.acme.io
    no quorum
    Unknown
Incident open Checkout is failing from 2 of 8 regions

Quorum met at 09:41 UTC. Frankfurt and London return 503; the other six regions still answer. Response times climbed for six minutes before the first failure.

Notified Slack #ops on-call email SMS webhook

Capabilities

Built around how
on-call actually goes.

Each capability exists because a team got paged for something avoidable. Here is what each one does and what it protects you from.

Checks & regions

Probe from eight places, agree before you page.

Every check runs from each compatible region you selected, exactly once per cycle. You set how many regions must fail before it counts as down, so one bad network path stays a footnote instead of an incident. Region choices adapt to the check type.

  • HTTP and HTTPS checks with status, redirect, and content-match rules
  • DNS A, AAAA, CNAME, MX, NS, and TXT expected-value assertions from all eight regions
  • Raw TCP-port checks from Finland and four Google Cloud regions, with an optional passive bounded banner rule
  • ICMP ping checks from Finland for hosts that do not speak HTTP
  • Per-check interval and per-check failure threshold
  • Response times kept per region, so slowdowns show up before failures

DNS and TCP checks are included with Complete and Ultimate.

US · IOWA
Iowa
104 ms
GB · LONDON
London
503
FR · PARIS
Paris
88 ms
DE · FRANKFURT
Frankfurt
503
FI · HELSINKI
Helsinki
71 ms
SG · SINGAPORE
Singapore
214 ms
TW · TAIPEI
Taipei
236 ms
AU · SYDNEY
Sydney
287 ms

Checkout · 2 of 8 regions failing · threshold met

Incidents

One incident per outage, opened and closed for you.

When the failure threshold is met, an incident opens on that cycle — not the next one. It carries the regions involved, the response history leading up to it, and who acknowledged it. When the service recovers, it closes itself.

  • Opens on the cycle the threshold is met, resolves automatically on recovery
  • Acknowledge to tell the rest of the team you have it
  • Notes on the incident for handover between shifts
  • No duplicate incidents when a service flaps
  1. Threshold met

    Frankfurt and London both return 503 in the same cycle.

    09:41
  2. Team notified

    Slack, on-call email, and SMS fire once — not once per region.

    09:41
  3. Acknowledged

    Dana takes it, so nobody else duplicates the work.

    09:44
  4. Resolved

    Both regions answer again; the incident closes itself.

    09:58

Incident timeline · checkout.acme.io

Domains & SSL

Certificates you never have to remember.

Add a domain and GeeksUptime checks its certificate daily. You hear about an approaching expiry with weeks to spare, and you hear about a fingerprint change the day it happens — which is often the first sign something was reconfigured without you.

  • Daily expiry checks with advance notice
  • Certificate change detection with a full history per domain
  • Issuer and validity window visible at a glance
  • Routed through the same notification groups as uptime
Domain
acme.io
Issuer
Let’s Encrypt R11
Expires
in 17 days
Fingerprint
changed 3 Jun

Certificate detail · acme.io

Maintenance windows

Planned work should not look like an outage.

Schedule a window for a whole workspace or just the checks you are touching, in your own timezone. Monitoring keeps recording results the entire time, so you still get the data — you just do not get woken up by your own deploy.

  • Scope a window to a workspace or to specific checks
  • Set it in your timezone; stored and evaluated in UTC
  • Results keep recording while incidents and alerts stay suppressed
  • Everything returns to normal automatically when the window ends
maintenance · production Active
  • Checkouthttps://checkout.acme.io
    timeout
    Suppressed
  • Public APIhttps://api.acme.io/health
    96 ms
    UP
Window open 22:00–23:30 Europe/Bucharest

Results are still being recorded. No incident opens and nothing is sent until 23:30.

Maintenance window · deploy night

Alert routing

Seven ways to be told,
scoped to who cares.

A notification group is a set of destinations plus the checks and domains it covers. Attach one group to your payment services and another to your marketing site, and neither team hears the other's noise.

Email Failure, recovery, and certificate notices to any number of recipients.
SMS For the incidents that cannot wait for someone to open a laptop.
Slack One incoming webhook per group, so each channel sees only its own services.
Telegram Your own bot token and chat ID per group for Markdown alerts in any chat or channel.
Mobile push Opt-in per group, delivered to everyone with access to that workspace.
HTTPS webhook Versioned JSON for your automation, runbooks, or a public status page.
AMQP queue Publish onto RabbitMQ when events need to fan out inside your own systems.

Under the hood

What happens on
every single cycle.

Worth knowing, because it is why you get fewer false alarms here than from a tool that pings once from one place.

Set up your first check
  1. Plan the run

    Every region the check is due from is written down before a single probe goes out, so nothing gets silently skipped.

    Plan
  2. Probe once per region

    Each region gets exactly one attempt. No hidden retries, which means no invented latency and no doubled load on your service.

    Probe
  3. Classify honestly

    A real response your rules reject is a failure. Our own timeout or transport error is not — that becomes unknown instead.

    Classify
  4. Decide the state

    Failures at or above your threshold means down. Every region definitive and below threshold means up. Anything else is unknown.

    Decide
  5. Act once

    Only a genuine change of state opens or closes an incident, and only then does anyone get notified.

    Act

Teams & reporting

Shared by default,
separated where it matters.

Workspaces

Group checks, domains, and alert routing by product, client, or environment. Switch between them without losing your place, and share a workspace with only the people who need it.

Permissions

Decide who can create checks, change routing, manage maintenance, or touch billing. Every change is written to the activity log with who did it and when.

Reports and exports

Availability by day, latency trends, and an SLA calendar per check. Export raw results as CSV or generate a PDF report when someone outside the team needs the numbers.

Get started

See it working on your own endpoints.

Thirty days of Growth features, no card, and your first check running in about two minutes.

No card · cancel by doing nothing · 30 days of Growth features