Changelog Roundup for October 12, 2026

The most important change this week is about deployment: you can now point your own domain at a site yourself, without the help of one of our admins, and the certificate is issued without a support request. For node operators, there is now a public status page that shows whether a problem is on our side.

For node operators

A public status page is now live at status.evolvingedge.ai. It checks the control plane, the portal, sign-in and our own sites once a minute, and shows the current state and the history of each since 4 October. It runs on a machine separate from the services it checks, so it keeps reporting when they are down. It checks from one location, so a problem shorter than a minute, or one that affects only some regions, may not show. Bookmark it if you run a node.

The control plane failed twice on 3 October. In the second outage, node check-ins, new registrations and node listings stopped answering from about 15:10 to 16:34 UTC. Two steps that run on every check-in could end up waiting for each other forever, and they now take their turns in the same order. We reproduced that cause in a test, but couldn’t confirm it from production. Nodes resumed checking in on their own, so you need to do nothing.

Node offline emails are now actually being sent. The emails we announced on 1 October were held instead of sent, because their wording was still marked as awaiting review. They have gone out since 4 October, so if your node went offline between 1 and 4 October and you heard nothing, this is why. Operators with older accounts are no longer skipped, because your sign-in email address is now used.

For customers

You can now claim your own domain in the portal and get its certificate automatically. Open a deployment target that has a custom domain, select Claim, and add the TXT and CNAME records it shows using your DNS management service of choice. Once the records are discovered, the certificate is automatically issued for you. An unverified claim lapses after 48 hours, and the same project then waits 48 hours before it can claim that domain again. A domain already serving your site needs nothing.

ee-cdn-cli now finds your output folder and keeps .git out of what it publishes. You can leave the folder off the command, turn on trailing-slash and clean-URL handling with a flag, pin the builder version for CI, and claim a custom domain with the new ee-domain command. Builds now leave out .git, .hg and .svn folders and stop on links that point outside the site folder. Until now, it was possible that both could be published with your site. Update and deploy as usual. If you have ever deployed a folder that held a .git directory or a link to a file outside it, deploy again with this release.

Every customer site timed out for about 20 minutes on 3 October, from about 13:40 to 13:59 UTC. Inactive browser connections had piled up on the control plane until there was no room for new ones. It now checks each connection every 30 seconds and closes any that stop answering. Sites kept loading through the second outage that day.

First requests to a site no longer wait several seconds. The first request after a gateway’s 30-second domain cache expired could take 3 to 13 seconds. The control plane now has dedicated capacity and does far less routine work. Response times should be improved but we did not test all customer sites, so tell us if yours is slower than you expect.

The new status page is the first place to look if a deploy fails, the portal won’t load or a site is slow. It doesn’t check your site, only our services and our own sites.

For everything else, see the full changelog.