Probably not many users who need the performance and can handle unexpected failover. There would also be the issue of setting the policy defaults effectively. Most users wouldn’t benefit from this footgun.

If you’re serious, you could probably automate this right now with your DNS provider and uptime monitoring.

I'm so serious that I already have failover after 1 hour at the registrar level, but those changes are not immediate and can take up to 24h to roll-in and roll-back due to DNS propagation and caching.

We seem to have hit the nesting maximum 1123581321, but to your point "There is no immediate option with DNS changes."

There's a huge difference in changing nameservers for a domain and simply changing host records.

This is a 7th level comment. Synu below[0] has a 9th level comment, and I've seen nested comments go quite a bit further.

[0] https://news.ycombinator.com/item?id=31821497

I think there is no nesting maximum (or if there is, it's much bigger than this). There's a limit which stops you replying to a comment immediately, to prevent super long quick-fire arguments.

Ah, sorry, misunderstood you. You can’t rely on them to change their host records when they’re down.

If you want CDN-independent automatic failover, look into anycast with two providers. If one of them is Cloudflare, use the tier that lets you manage your DNS elsewhere.

There is no immediate option with DNS changes. CF can’t immediately remove their IP from the route. Sounds like you’ve solved your problem in the sense that you have an automatic failover, though, which is good.

Typically CF _is_ the DNS provider though, right?

Only up until the $200/mo tier. This kind of feature would be locked at that level anyway.

As I said to the OP elsewhere, he should be doing something like anycast to multiple CDNs if this is critical.