Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I was hoping there would be a discussion of this here with some inside baseball, but then HN was down.


The Cloudflare blog suggests it's BGP problems: https://blog.cloudflare.com/cloudflares-view-of-the-rogers-c...


this is pretty far outside my area of expertise so maybe i'm missing something, but isn't the cloudflare blog just describing what they're seeing from the perspective of somebody who monitors BGP traffic, rather than assigning the blame to BGP?

it seems like the BGP traffic signature of any isp-level outage, regardless of the cause, would look like this. they went down, so they stopped advertising that they could accept traffic. then they tried to accept a little bit of traffic, but that didn't work so they turned it off again.


Yes, exactly. BGP withdrawal is likely a symptom, not the cause.


How does that relate to the Rogers cell service outage, though? Surely SMS doesn't require BGP?

It feels like there's a big, fundamental problem affecting a variety of Rogers infra, where BGP and cell problems are just symptoms we can observe. CloudFlare says it looks more like a failure than an attack; I'm not entirely convinced.


I don’t think it’s an attack.

I’m in Canada at the moment with my UK SIM and using Rogers internet just fine. Completely unaware of any problems until my friends warned me about needing cash and I then double checked and still, phone internet via Rogers on a roaming SIM worked fine.

My guess is their voice, sms etc all are IP these days which was impacted by their BGP issues. Whereas my UK carrier roaming gives me a UK IP still, unrelated to Rogers advertised blocks.


> My guess is their voice, sms etc all are IP these days which was impacted by their BGP issues.

Impacted, yes. But "no service"? That's dubious.

I hear you on foreign SIM working. I got online at a cafe just long enough to buy a Yessim eSIM. When I activated it and turned on data roaming, I had LTE working on... the Rogers network! That I didn't expect.


> Impacted, yes. But "no service"? That's dubious.

Nope, in contrary I will fully believe it. Circuit-switched networks are dead, period. Most "circuit-switched" networks are actually IP networks just with custom FEC'd and prioritised protocols so that there's backwards compatibility. Most Subscriber Authentication systems since UMTS and 3G CDMA (except for super-early deployments, and since they have LTE they would have migrated it if it were the case) have moved into an IP-based system because it's cheaper, and if that's down then everything is down.


It's always BGP......


Sometimes it's DNS...


Should just merge this all into systemd-bgp-dns; kill all the birds with one stone.


Well, Pottering quit RedHat to work for MS, maybe that's exactly the plan


Unfortunately BGP problems make DNS problems look trivial :(


I was so confused when I couldn't access HN - I thought to myself maybe HN was hosted on Rogers network or something.


There's a Wikipedia article, though it's still sparse. It should have post mortem information once that's official.

https://en.wikipedia.org/wiki/July_2022_Rogers_Communication...


Ditto, I expected to this be front-page on HN with many upvotes.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: