Sorry about this, I still don't have any answer or explanation from Vultr or CloudFlare at this point. Most likely cause IMO is that CloudFlare (accidentally?) blocked one or many big ranges of IPs belonging to Vultr (and maybe some other providers as people seems to say Vultr was not the only impacted). I noticed during the incident this morning for example that I could ping CloudFlare IPv6 (ICMP) but not connect through TCP (port 443). So this sounds more like a firewall than a routing issue from what I could see.
I'll update once I have anything else in https://status.updown.io/issue/1e196616-1368-43a0-8c04-82cff.... For the moment I'm keeping the mitigation in place just in case.
If you have more details about this from CloudFlare or elsewhere I'll be happy to hear it :)
So this confirms what I suggested above, I suppose they choose not to respond with an HTML page here because it would generate too much traffic, and maybe it was a lower level TCP attack.
This also probably explains why Vultr doesn't want to answer me if they were "responsible" for the DDoS attack that got them blocked.
Same thing with uptown.io some issue with vultr to cloudflare triggering false positives...unless you're connecting from vultr of course