Website uptime monitoring with the cause in every alert

When a client's site goes down, the alert tells you what failed and which locations can still reach it, so you can tell a dead server from a blocked request before you pick up the phone.

How sites fail without looking like a crash

A site can stop working in ways nobody calls an outage. A certificate expires and every browser refuses the page. A host starts timing out under load. A firewall rule returns a 403 to some visitors, or a deploy turns the homepage into a 500. A site can also keep answering, only slowly enough that visitors give up. In our experience the client finds out first, from a customer.

What OctoPing checks

OctoPing requests the site's URL, follows redirects, and reads the status code of the final response. A status from 200 to 399 counts as up. Anything else counts as a failure: a 4xx or 5xx status, a timeout, a failed DNS lookup, a refused connection or a TLS error. Each failure is classified, so the alert says Timeout or DNS resolution failed instead of a bare error code.

The default is a check every 5 minutes, and you can choose every 30 seconds, every 1 minute, every 2 minutes, every 5 minutes, every 30 minutes or every hour for each site. The fastest interval depends on the plan:

The response time is the time until the first response headers arrive, so a large page doesn't read as a slow one. A site protected by HTTP basic auth can still be checked: you add the credentials, and any custom request headers, to the site, and the password is stored encrypted.

Degraded: up, but slow

Set a response-time threshold in milliseconds on a site, and OctoPing marks it Degraded when the latest response is slower than that while the status is still up. The threshold is off until you set it. The first slow response sends a slow-response alert, and the alert repeats at the rule's repeat interval while the site stays slow. When the response time drops back under the threshold the status clears quietly, with no recovery message.

When it alerts

Each alert rule decides how many failed checks in a row it takes to send the first down alert (3 unless you change it, from 1 to 10) and how many minutes pass between reminders while the site stays down (10 unless you change it). When the site comes back, the recovery alert gives the downtime and the cause. An outage runs from the first failing check to the first one that succeeds, so a two-hour outage with reminders counts as one outage.

The down alert lists every location's result. Before it goes out, OctoPing compares the failure with the regional probes (see multi-region monitoring). Channels, switches and the retry settings are on the alerts page.

Planned work: maintenance windows

A maintenance window tells OctoPing that a site or heartbeat is expected to be unavailable. You give it a name, the sites and heartbeats it covers, a time zone and a schedule, either once or weekly on the days you pick. Members can schedule windows up to 24 hours long, and Admins and Owners up to 7 days.

• Paused. Down alerts and their reminders, slow-response alerts, status-change alerts and missed heartbeats stay quiet for the covered sites while the window runs. The site shows a blue Maintenance status.
• Kept out of the numbers. Checks made during the window don't count toward uptime and aren't counted as downtime.
• Still sent. Certificate and domain expiry warnings and change alerts for content, DNS, robots.txt and hosting still go out, because an expiring certificate isn't planned work.
• Outages already open. If a site was already down and alerted before the window began, that outage stays open and its reminders pause. Its recovery alert goes out as soon as the site recovers, even inside the window.

A site that is still failing when the window ends alerts under the normal rules from its first failing check afterwards. Published status pages show a window that's running or starts within the next 7 days under Scheduled maintenance, for the sites and heartbeats on that page. The window's name is shown there, so write it for your client.

How it fits the price

Uptime is checked from all four vantage points, and every site can also use the SSL, domain, DNS, hosting, robots.txt and content checks. No plan limits which checks a site can use. Maintenance windows are on every plan. Plans start at $19 a month for 20 sites. See pricing for the full list.

Every account starts with a 14-day free trial. No card required. Start your 14-day trial and add a client's site to see it working.