What online, degraded, and offline mean on a site tile

The short answer

On a DashCanopy tile, online means the last health check received a successful response in under five seconds. Degraded means the site responded successfully, but the check took five seconds or longer. Offline means the check failed: no response, or an error status. The three words are about the last check, not about a promise that the site will stay that way.

That definition is enough to stop the most common misread. A yellow or degraded tile is not “down.” A green tile is not “fast for every visitor in every country.” An offline tile is a failed check from the dashboard’s checker, which you still confirm by opening the URL yourself before you restart a server.

What the checker is actually doing

DashCanopy asks the URL you saved for that site whether it will answer. The request is a health check, on a schedule set by the plan. Starter checks every 10 minutes, Pro every 5 minutes, and Agency every 1 minute. Between checks, the tile keeps the last result. If you deploy at 9:01 and the next Starter check is at 9:08, the tile can stay green, or stay red, until that next check returns.

A response in the success range, with a measured time under five seconds, is online. The five-second line is deliberate. Some hosts take a couple of seconds to wake on the first request after a quiet period. Treating every cold start as an outage would train you to ignore the tile. Treating a five-second success as degraded tells you the site answered and that the answer was slow enough to notice.

A response that never arrives, times out, or comes back as an error status is offline. HTTP status codes are standardized; a 500 from the app and a connection failure are both failed checks from the tile’s point of view, even though the fix is different. The status code list on MDN is the reference for what those numbers mean. The tile’s job is to surface the failure. Your host’s logs tell you which failure it was.

How to react to each state

When the tile is online, leave it alone during a health pass. Online is not a performance report, a search ranking, or a revenue figure. It means the last check succeeded quickly enough to count as healthy. Read visitor and revenue stats as a separate step, the way the morning check lays out.

When the tile is degraded, open the site once in a browser. If the first load is slow and a refresh is normal, you are probably seeing a cold start or a heavy page, and the next check may return to online without any action. If every load is slow, the host is the next stop: CPU, a stuck deploy, an upstream the page waits on. Degraded is a reason to look, not a reason to announce an outage. Say what you observed: the check succeeded, and it was slow.

When the tile is offline, try the public URL yourself before you change anything. Three outcomes are common.

You cannot load it either. Use the host’s console, the last deploy, and DNS. The dashboard has named the site. It does not SSH in for you.

You can load it, and the tile stays offline on the next check. Compare the URL stored on the site with the URL that works. A missing slash, an old http address, a hostname the firewall blocks, or a checker that is not logged in can fail a check against a site customers open fine. Fix the monitored URL or the block before you treat it as downtime.

The tile recovers on the next check. Record the window if it mattered, and continue the morning. Intermittent failures are still information. They are not a reason to rebuild the site because one check missed.

Response time is not uptime, and neither one is analytics

People fold these together because they all live on a “is the site okay?” list. They answer different questions.

Uptime, in this dashboard, is whether checks succeed. A site can pass every check and still be the wrong page, a maintenance notice that returns 200, or a checkout button that does nothing. The health check does not click through a purchase.

Response time, here, is how long that check took. It is not a lab score of Largest Contentful Paint, and DashCanopy is not a Core Web Vitals product. Google’s page experience documentation is about what search measures in a browser (page experience). Use that when the question is search. Use the tile when the question is whether your checker got an answer, and how long it waited.

Visitor and revenue stats are a third question. They show up when the site reports them. A tile can be online with empty stats because nothing is pushing numbers yet. That is a setup gap, described in how stats land on a tile. Do not relabel it as degraded.

How often the words can change

On Agency, a check every minute means a recovered site can flip back to online while you are still reading the host logs. On Starter, ten minutes can pass between a failure and the tile you see. Neither interval is a paging policy by itself. Email notifications are included on Starter, Pro, and Agency. SMS, if you want a text, is on Pro and Agency and uses your own Twilio account. The tile is the record. The alert is how you hear about a change when you are not looking at the record. Choosing between them is covered in email alerts versus SMS.

If you need a tighter interval than your plan, that is a plan question, not a settings trick. The intervals above are the ones on the pricing page. Starter is $19 per month, Pro is $69, and Agency is $149. Annual prices are $190, $690, and $1,490.

Limits worth remembering

A single check is one request from one place. It can miss a failure that only happens for visitors in another region, on a phone, or behind a login. It can also fail when visitors are fine. Read the tile as the checker’s last result, then confirm with the URL and the host when the result is degraded or offline.

The tile will not tell you which line of code broke. It will tell you which site to open first when you have more than one.

Questions about tile status

Does degraded mean I should take the site down for maintenance?

No. Degraded means the last successful check was slow. Confirm it in a browser. Take the site down only if you already have a maintenance reason of your own. The status word is not an instruction to enable a maintenance page.

Can a site be online while customers see an error?

Yes. If the URL you monitor returns a success code, the tile can be online while a deeper path, a logged-in screen, or a third-party script fails. Monitor the URL that best represents “this property answers,” and use logs or the product itself for paths the checker does not request.

Why did the status change when I did not deploy?

Hosts sleep, certificates renew, DNS changes, and upstreams blip. A status change without a deploy is still a real check result. Look at the clock on the check and the host’s own status page before you assume a bad release.