Just like AWS, then. "Some users are experiencing increased error rates" = "Everything has been down for hours"

"Everything is fine, unless you're Carl. There's a massage outage, but only at Carl's house. Sorry, Carl."

I'm also experiencing a massage outage. Please send masseuse.

Goddammit some (most?) days I can't type. "Massive"

I got the missive. Thanks.

I remember when S3 was down and the status was green because the updates for the status page with pushed via S3.

That's not just ironic, that's stupid. How do you count on S3 to update S3 status? Isn't that a huge design fault?

Yes and they fixed that. :)

Azure too. During the most recent outage a couple weeks ago their Twitter account acknowledged the incident an hour before the status page did.

So no matter where you go for your cloud services, you're guaranteed a useless status page. Yippee.

AWS is no better. Something from 2015 I remember: https://twitter.com/SIGKILL/status/630684777813684224?s=19

I swear most status pages are run by folks who aren't "there".

It’s an easy problem to fix as basically services should emit performance data, openly, and the status page should just summarize that. So if a service doesn’t report out, it’s assumed down or erroring out.

Having an excel file where people enter statuses is not very useful to me as a customer. That’s more like a blog.

I haven’t written a status page in a while, but the rest of my infrastructure starts freaking out if it hasn’t heard from a service in a while. Why doesn’t their status page have at least a warning about things not looking good?

In my experience public status pages are "political" and no matter how they start tend to trend towards higher management control in some way... that leads to people who don't know, aren't in the thick of it, don't understand it, and / or are cautious to the point that it stops being useful.

Not only political, but with SLAs on the line they have significant financial and legal consequences as well. Most managers are probably happier keeping the ‘incident declaring power’ in as few a hands as possible to make sure those penalty clauses aren’t ever un-necessarily triggered.

That’s fraud in other industries.

Same with most corporate twitter feeds. I’d like to follow my public transit/airport/highway authority, but it’ll be 10 posts about Kelly’s great work in accounting for every service disruption.

And No, I don’t want to install a separate app to get push notifications about service disruptions for every service I use.

A good Twitter account is a wonderful thing....the bad ones hurt so bad.

Ugh. I guess that just goes to show that any metric can be politicized.

It's just Goodhart's law in effect: If a status page is used as a target metric in an SLA the status page ceases to be a useful measure.

Status pages are the progress bars of the cloud.

I worked on the networking side for years.

Now the web development side and I'm all "Wait a minute...are there any progress bars that are based on, anything real!?!?!"

I should have known...

Digitalocean has the same issue: status pages are actually manually updated and no live data is fed into them.

Was noticing massive issues earlier and thought that maybe my account was blocked due to breaching from TOS as I was heavily playing with Cloud Run. Then I noticed gitlab was also acting up but my Chinese internet was still surprisingly responsive. Tried the status page which said everything was fine and searched Twitter for "google cloud" and also found nobody talking about it. Typically Twitter is the single source of truth for service outages as people start talking about it

I think this might be a static page they are hosting on Akamai?

Doesn't look like it:

https://www.whoishostingthis.com/#search=status.cloud.google...

They update the page manually.