Uptime checks

A monitor checks one thing on a schedule from several places at once. It only counts as down when those places agree, so one location's bad network minute never wakes anyone up.

5 min read

Creating a monitor

Open Argus → Uptime and choose Create monitor.

Kind	Target	Checks
HTTP	A full URL, like `https://storefront.ng/health`	The response status, and optionally a keyword in the body
TCP	A host and port, like `db.storefront.ng:5432`	That the port accepts a connection
DNS	A domain name	That its `A` record resolves, and optionally that the answer contains a value
TLS	A host (port 443 unless you give one)	The certificate, and how many days it has left
Ping	A host	That the host answers on port 443, then 80

Ping monitors connect over TCP rather than sending ICMP, because ICMP needs privileges the checkers don't have.

Targets must be public. Private, loopback and link-local addresses are refused when you save the monitor, and again when it is checked, so a hostname that resolves to a private address is refused too.

What passes

HTTP. Any 2xx or 3xx status passes. Add a keyword and the first 64 KB of the body must contain it as well.
DNS. The domain's A record must resolve. Add an expected value and the answer must contain it, such as an IP address.

Slow and expiring

A check that takes longer than 3 seconds is slow. A TLS certificate with fewer than 14 days left is flagged too. Both show as degraded, in gold, rather than down.

How often, and from where

Checks run from three locations:

Location	City
Nigeria	Abuja
Europe	Nuremberg
United States	New York

Every monitor uses all three unless you pick fewer when you create it.

A monitor runs every 30 seconds, 1 minute, 3 minutes or 5 minutes. Your plan sets the shortest interval you can choose (see Argus → Usage & plan).

When a monitor counts as down

Each location reports its latest reading. A reading counts while it is fresh, meaning newer than two intervals plus a minute. Then the locations vote:

The monitor is down only when a strict majority of the fresh locations, and never fewer than two, report it down.
If fewer agree, the monitor is degraded and the list says where, such as Down in EU. That shows in gold and pages nobody.
If only one location has a fresh reading, it needs two failures in a row before it can call the monitor down, because there is nobody to agree with it.

Uptime percentages

Uptime is measured from confirmed downtime, not from raw readings: a minute counts as down only when the locations agreed it was. A location that alone sees the monitor down doesn't lower the figure, which is why a monitor can read Down in EU and 100.00% at once. Argus marks that case Not counted yet.

Percentages are rounded down to two decimals, so a day with any confirmed downtime never shows 100%. Days are counted in WAT, and time before the monitor existed counts neither way.

When it goes down

When a monitor turns down, Argus opens an incident and tells your workspace's alert recipients by email, and every channel that is on, such as Slack or a webhook: once when it goes down, and once when it is back up. If nobody acknowledges the incident with I'm on it, Argus sends one reminder after 15 minutes.

A monitor's page shows response times from each location, its daily uptime and its incidents. Pause a monitor to stop checking it without deleting its history.

Check a site once, without an account

The free uptime check tests any URL once from all three locations and gives you a shareable result. No sign-up.