Hexnode portal not loading with ERR_CONNECTION_REFUSED and Windows enrollment network errorSolved

This behavior is most likely related to portal availability or latency rather than a device-side enrollment configuration issue. A few indicators point to that:

• The portal does not load across multiple browsers, devices, and networks.
• Another technician is affected as well.
• ERR_CONNECTION_REFUSED means the connection reached the destination but was refused by the server or service endpoint.
• ERR_CONNECTION_TIMED_OUT means the browser did not receive a response in time.
• The Windows enrollment agent depends on the portal being reachable, so it can show A network error has occurred if the portal cannot be contacted during setup.

In this situation, browser cache clearing or reinstalling the Windows enrollment agent usually will not resolve the issue. The recommended checks are:

1. Confirm whether the issue affects more than one technician.
2. Test from a separate browser or incognito window.
3. Test from a different network.
4. If all tests fail with connection refused or timeout errors, wait for the portal connection to recover or raise it as a portal availability issue.

Once portal access is restored, pending Windows enrollments may continue automatically or complete when the enrollment agent is run again.

That lines up with what I saw. One browser tab eventually logged in and a pending enrollment seemed to go through, but new sessions in incognito and on another device were still timing out. Submenus also would not load at first.

That can happen during partial recovery. An existing session may load before new login attempts or menu navigation fully stabilizes. It is best to verify recovery by checking all of the following:

• The login page opens in a new private/incognito session.
• The dashboard loads without timing out.
• Device, policy, and support pages open normally.
• The Windows enrollment agent can complete setup without the network error.

If only an existing browser tab works while new sessions still fail, the portal should still be treated as not fully reachable.