4 ms·
CloudFlare is a good company and everyone has outages. IMHO the post-mortems they post are not only some of the best I've read from a big company, but they are
by redm 6y ago
CloudFlare is a good company and everyone has outages. IMHO the post-mortems they post are not only some of the best I've read from a big company, but they are produced quickly.
I only wish they could update cloudflarestatus.com more quickly. Shouldn't there be some mechanism to update that immediately when there is an incident? When the entire internet knows your down and your status page says All Systems GO! it looks very poorly on them.
- aeyes 6y agoTheir comment regarding the late update here: https://news.ycombinator.com/item?id=23878496 https://news.ycombinator.com/item?id=23878496 > Because of the way we securely connect to StatusPage.io from most locations where our team is based. The traffic got blackholed in ATL, keeping us from updating it.
- crest 6y agoThey could add a deadman switch to their status page.
- ithkuil 6y agoA meta status page
- marksomnian 6y agoSoon you'll need status pages for the status pages.
- fernandotakai 6y agoalready exists ;) https://metastatuspage.com/ https://metastatuspage.com/ (it's statuspage.io's status page)
- nostrebored 6y agoCircular dependencies like this are not a good look when the core of your business is networking...
- aeyes 6y agoYep, I was joking with our rep that them using Cloudflare Access for their internal services sounds like a problem waiting to happen. Guess I wasn't wrong, they might even have lost access to internal monitoring systems which is pretty unfortunate in such a situation. If you ask them about Cloudflare Access they will happily tell you that it was built for internal tool access and that they use it for everything, later they went on to sell it as a product.
- dylan604 6y agoAn example of dogfooding your own product is actually a bad idea?
- microcolonel 6y agoWhile they could dogfood the product, status monitoring systems should be separated from your bread-and-butter product's failures. If you are in the business of messing with BGP, then the BGP that controls the routes that let you report outages should not be the same one you are messing with regularly, or at least, there should be redundancy.
- kochthesecond 6y agoWhen Google Cloud went down a few years ago, they were unable to access internal monitoring because the bad bgp change overloaded their networks if i remember correctly.
- earthboundkid 6y agoIsn’t it status page 101 that you host it in the competition, so it stays up when you’re down? :-/