Custom domain at Cloudflare Pages is stuck in 522 error

Before, I used to have a Cloudflare Pages website (let’s call it website A) with a custom domain (hosted by CF) pointing to it. No problem.

Then, I created a new Github repo for a totally new version of my website and deployed it with Cloudflare Pages (website B). Website B works perfectly on the Cloudflare Pages .dev domain, so I deleted the custom domain in the settings of website A, and pointed it to the new website B. The resulting DNS seems fine, with CNAME and such.

But… in the Cloudflare Pages console of website B, the domain is stuck in “Inactive (error)”, resulting in what apparently is a 522 error. So far, Cloudflare support hasn’t been able to help in a response to my ticket (I seem to be getting canned responses only for the past days).

Website B has been down for three days. Help would be greatly appreciated!

1 Like

Can you please post the ticket #?

#2142348

Does the pages project name contain punycode by any chance?

The custom domain is a .io name. The Pages dev domain is formatted as domain-io.pages.dev

Have you tried removing the DNS record, leaving it for a while then re-adding it again?

Yes, I removed and added it several times - and the DNS seems fine. The Pages settings don’t work. The 522 error occurs when the source does not respond (i.e. a server is not available or so). The Pages website just does not want to accept the custom domain (it sits in an error state in the Pages console).

And it works if you change the custom domain back to point at Website A (if it still exists)?

Let’s try that. Just a sec.

I deleted the custom domain from website B. Then, I added it to website A (where it originally was). For website A, it now also immediately jumps into “Inactive(error)”.

By the way, I didn’t want to do this test because I did not want to confuse support. However, since Monday, nobody really looked at it, so… It does worry me, though - since we were planning on building this out as a production project. How do you guys deal with support?

If you join the discord server here:

You can ask the workers/pages team directly (so you may well get a quicker answer) :slight_smile:

I’m not sure what else could be causing the problem for you though.

1 Like

Thanks, will try.

Apparently, the domain was flagged and banned when I switched the custom domain to another Pages project - according to the dev team on Discord. They don’t know why or how to remove that ban. As per their advice, I reached out to [email protected]. No answer so far.

In the meantime, I also received another message from Support, saying that a “DS record” may be the culprit. But they did not explain how to get rid of it or how to fix it. So, I am still stuck.

Update: our website has been down for a week now and all we hear are crickets. What we know is that the domain was flagged and banned during the switch between Github repos. I don’t know why. So far, Customer Support has been silent since the (very helpful) Pages dev team raised the issue, as is the [email protected] team. Moving the website to Workers entirely does not help, since it is the flagged domain that is blocking everything - and moving the entire DNS would be impossible, because our API is based on Workers. So, we are entirely blocked.

Update: after an intervention of the Pages dev team, the ban has been lifted and we have a live website again! Hurray for the team!

1 Like

Yikes this is worrying, infact frightening, this level of inaction for days despite repeated attempts to reach support via officially suggested channels and in the end you had to use an unofficial channel and wait several days for a mistake that appears solely on Cloudflare’s shoulder.

When DigitalOcean did this the community was mortified ("DigitalOcean Killed Our Company" | Hacker News), what followed was a complete review of abuse procedures, a promise of more thorough review (and second vetted) before any disabling/suspending action, a complete communication revamp, and a direct and public apology from the CEO.

Do you still feel comfortable with Cloudflare after this? Has there been a postmortem on this disaster?

Whilst this may of been a hobby project many of us are running businesses (on paid plans) and if [email protected] is silent then we’re potentially screwed.

Have you consider a HN post? Not to gaslight CF for this terrible response but to ensure there’s no future occurrence for yourself and others, people like John Graham-Cumming (CTO) respond promptly on HN.

salepurchaseplot, eager to understand, was this a cloudflare problem like the OP experienced - that could only be fixed by Cloudflare staff, or was it something that you resolved yourself eventually aided by community help and/or documentation?