[0xffany]

Everything appears to be down as of 18:43 UTC... https://downdetector.com/

[voytec]

Yeah. This service was presenting charts likely probed from inside GCP. I was on a call with a Google rep, someone pointed out that "AWS is also down" and I foolishly said something about "possible BGP attack" out of spite, before checking AWS availability myself. Shame on me.

[toast0]

Didn't have the feeling of a BGP issue, most services I was working with were reasonably quickly returning failures, as opposed to lingering death.

[yard2010]

I love this kind of fake news. It's like that scene from Scary Movie (can't remember which one) in which someone says "I heard the japs took out one in Kikoman" :')

[AlienRobot]

Wait, it's all Google?

[deepsun]

Google was the first to report probably.

[bananapub]

all cloud

[plateng000]

"always has been"

[sillypuddy]

Well that's interesting. I wouldn't expect AWS or Microsoft 365 to be affected by a Google outage.

[paxys]

Who said it's a Google outage?

[AdamJacobMuller]

Google. https://status.cloud.google.com/regional/americas

[paxys]

It's more likely to be a broader issue that is affecting AWS, Microsoft, Cloudflare, GCP. They aren't all dependent on Google infra.

[ikiris]

Oh look, they were.

Cloud flare was really the gcp problem. Most of the others are going to be dependencies on cf or random Google stuff.

Discord for example was gcs for updates, etc

[patapong]

Perhaps their detection logic is running on Google cloud /s

[throitallaway]

I believe Downdetector displays user reports.

[brentm]

Yea I am pretty sure that if you're checking if a service is down your essentially casting a vote that indicates that service is down.

[lysace]

Kind of a missed opportunity for Ookla - who's running both downdetector.com and speedtest.net.

The have software running in most ISPs around the world:

https://help.speedtest.net/hc/en-us/articles/360039164793-Ho...

(OTOH, it's not always trivial to define/detect an outage.)

[peanut-walrus]

Downdetector in incidents like this is 100% misinformation.

[johanyc]

Why

[peanut-walrus]

Downdetector does not actually monitor the services. It aggregates user reports from socials etc. For large-scale incidents, the reports get really noisy and it will show that basically everything is down.

[boneitis]

I thought that was the whole premise of Downdetector, no? User reports, because first-party status updates are tightly controlled by those first parties?

Was not basically everything (hyperbolically speaking, of course) practically impacted today?

How much weight really comes from those social media posts? Is there an indirect effect of people reading these posts, then flocking to hit the report button, sight unseen?

[baobun]

Who watches the watchmen?

(downdetector infra also likely affected)