Down Detector

Is it the service, or your connection? Let's check.

Services at a glance

How to use Down Detector

  1. Search for a service such as Codex or Discord, or enter a public website address.
  2. Choose a matching service to see its official status, affected components and available incidents.
  3. For any public website, run a reachability check and compare its HTTP and DNS evidence with what you see in your browser.
  4. Check the source and timestamps before sharing a result. Automatic refresh updates official feeds while this page stays open.

Two checks, two different questions

An official status feed tells us what the company has acknowledged about its services. A website check tells us whether one public URL answered an external request. We keep these signals separate: a homepage can work during a login incident, and a security challenge can block our check while the app works for you.

What does each service status mean?

Operational

The connected provider feed lists the service as operational. An individual account, feature or region can still have a problem.

If it still fails for you, compare another network and device before changing your settings.

Degraded performance

The provider reports reduced performance, such as slower requests or intermittent errors.

Check the affected component and incident notes. Avoid repeatedly submitting payments or other important requests.

Partial outage

Part of the service is reported unavailable; other components may still work.

Match the affected component to the feature you need, then follow the provider's updates.

Major outage

The feed reports a major outage within the service or components we cover, not necessarily every product the company offers.

Use an alternative if the task is urgent. Check the official incident for recovery updates.

Maintenance

The provider has marked a component or service as under maintenance.

Read the maintenance notice for its scope and schedule. A planned end time is not a recovery guarantee.

Incident reported

An active incident is listed, but the available feed may not specify its full severity.

Open the official incident to see which features are affected and what the provider recommends.

No active incident listed

The connected incident feed lists no active incident. This is different from a successful test of every feature.

Compare the result with the feature that is failing. A new issue may not have been acknowledged yet.

Status unavailable

We could not verify the feed, or it does not provide enough information to classify the status.

Open the official status page and try Refresh later. Unavailable data does not mean the service is down or working.

Still having trouble? Start with your symptom

It works on mobile data, but not Wi-Fi

Try another device on the same Wi-Fi. If both fail, compare your internet provider's local outage checker. A VPN, DNS setting or network restriction may also explain the difference; it does not by itself prove a router fault.

The homepage opens, but login or one feature fails

Look for the matching component, such as authentication, messaging or downloads. A reachable homepage cannot confirm those features. Record the error and check the provider's incident before repeatedly retrying.

Only one browser or device has the problem

Compare a private window or another browser. Extensions, stored browser data or an app version may be involved. Save unsent work before clearing data; clearing cookies can sign you out.

It fails on different devices and networks

Check official incidents for the affected feature and region. For broadband or mobile service, use your provider's address-based checker. Several failed tests are useful evidence, but do not establish a worldwide outage.

Before contacting support, keep the time, affected feature, network and exact error. Share a public status link or downloaded report, never passwords or private account links.

How to read the status charts

Components are not people

The health bar counts components returned by the provider, such as APIs or regions. It does not measure customers affected, traffic or uptime. A large number of healthy components can coexist with a problem in the one you use.

Incident starts are not outage duration

Where history is available, the chart groups returned incidents by their start date over the last 30 UTC days. An incident lasting several days is not a new incident each day. A zero or missing history is not proof of uninterrupted service.

The directory is a feed snapshot

The overview counts service signals, not visitor reports. Related products can share an underlying provider incident, so do not add them up as independent outages. Unknown feeds stay separate from services with no issue listed.

How fresh are these results?

Fetched and provider updated are different clocks

Fetched is when ClockTools obtained a successful source response. Provider updated is the timestamp supplied by that source, when available. It is not necessarily the incident start or resolution time, and an old provider timestamp alone does not prove an outage.

Refresh checks again, but does not bypass every cache

Official feeds refresh about every two minutes while this page is visible and auto-refresh is enabled. Successful status responses may be reused for two minutes, and separate history responses for five minutes. Refresh can therefore return the same timestamps. A website check is a one-time request, not continuous monitoring.

Reports can lag behind the problem

A provider may publish an incident after users first notice it. If a source cannot be verified, we show unavailable data rather than an all-clear. Compare your symptoms, the checking location and the official source before deciding how widely a problem is affecting people.

What our status directory covers

We read public provider feeds for AI services, collaboration apps, gaming platforms and cloud infrastructure. Google Workspace and Google Cloud supply incident feeds rather than proof that every individual account works. A Codex incident is not automatically a ChatGPT incident. Component filters keep their results separate, and unmapped or unavailable data is labelled honestly.

Your checks and privacy

Official feed requests do not need your account or API key. Website addresses are sent to our server to perform the external check; only public HTTP and HTTPS addresses are supported. Use ordinary public links, never private dashboards or URLs containing secrets. We do not save your search history in this tool. Public provider responses are cached briefly to reduce repeated requests.

History without invented uptime

The incident list contains only records returned by the official source. Feed retention differs, and some providers expose no history through the connected endpoint. We do not turn missing incidents into a 100% uptime claim, fabricate visitor reports or estimate how many people are affected. This page is independent of Downdetector and the services listed.

Checking Verizon, Spectrum, Virgin Media and Facebook

An internet outage is often local. Choose Verizon, Spectrum or Virgin Media for the provider's own area or account checker. For Facebook, compare a public website check with the feature that is failing on your device. Roblox has an official status feed; Steam offers a narrower public API signal. These different sources should not be treated as the same kind of outage measurement.

Down Detector questions

Can I check any website?

You can request a one-time check of a public HTTP or HTTPS website, including sites outside the official directory. Private addresses, unusual ports and credential-bearing URLs are rejected. Firewalls and bot protection may prevent a conclusive result.

How live are the statuses?

Official feeds refresh every two minutes while the page is open, with a short shared cache. Provider announcements can arrive after an incident starts. The fetched time and provider update time mean different things; neither is a guarantee that every region or account is working.

Can I check whether Instagram is down?

Yes, search Instagram to check its public website. We do not have a verified official feed covering the whole consumer app, so a responding page does not confirm that Reels, direct messages or login are working.

Does a 403 or 404 mean a global outage?

No. A 403 usually means the server refused the request; a 404 means that particular page was not found. Both show a responding server. Even a timeout or 5xx result from one checking location needs more evidence before calling it a worldwide outage.

Is this tool free?

Yes. No login, installation or personal API key is required. ClockTools reads public status feeds and performs bounded website checks. It does not access your private accounts or run paid AI prompts to determine status.

Does ClockTools collect public outage reports?

No. These results use connected provider feeds and individual website checks, not crowdsourced reports. We do not display community report totals, affected-user estimates or a crowdsourced outage map. ClockTools is independent of Downdetector.

Can you tell me when a service will be back?

Only the provider can give an authoritative recovery update. Follow the linked incident or maintenance notice. We do not invent restoration times or treat a scheduled maintenance end as a guarantee.

Website Down Checker Internet Speed Test