Hacker News Reader

Comment by suby | Hacker News Reader

I was doing development work which uses a server I've got hosted on digital ocean. I started getting intermittent responses which I thought weird as I hadn't changed anything on the server. I spent a good ten minutes trying to debug the issue before searching for something on duckduckgo, which also didn't respond. Cloudfare shouldn't be involved at all with my little site, so I don't think it's limited to just them.

View on HN

Yeah, something happened to ipv4 traffic worldwide. Don't see how that could happen.

Let me guess: somebody misconfigured BGP again?

likely

https://puck.nether.net/pipermail/outages/2020-August/013198...

That's definitely going to be an interesting postmortem.

Seconding this. Had some ssh connections timing out repeatedly just a bit ago. Also got disconnected on IRC.

IKEA had their payment system go down worldwide also. I really doubt that uses Cloudflare.

It's not a just CloudFlare outage, its a global CenturyLink/Level3 outage

Is there a ranking board for which carriers have caused the most accumulated network carnage out there? I think the world deserves this.

Me too. I can only connect to one of my DO servers. The rest are all unreachable.

As noticed in another comment I see loads of problems within Cogentco, all on *.atlas.cogentco.com. Might the problem lies there?

Cogent and Cox are also having problems, but we are seeing a lot more successful traffic on Cogent than CenturyLink. It appears that CL is also not withdrawing stale routes. It seems CLs issues are causing issues on/with everything connected to it.

Same here. I actually opened a support ticket with them because I was worried my ISP had started blocking their IP addresses for some unknown reason. Luckily it seems to clear up, and in the ticket they mentioned routing traffic away from the problematic infrastructure. Seems to have worked for now for my things.

Yup, definitely noticed earlier outages to both EU sites and also to HN. Looked far upstream because many sites/lots of things worked fine. Good to see it's at least largely fixed

I had problems accessing my Hetzner VPS', but I haven't tried connecting directly with the IP. So I suppose it could be a DNS thing?

Reply on news.ycombinator.com