If you have S3 bucket logging turned on, the cost will continue to grow forever, because it's literally generating its own transaction charges and then snowballing the storage. It's like an evil genius's perpetual motion machine. You'd think they'd know how to delete all of those things when the account is closed, since they certainly know how to bill for them.

I once closed or deleted our s3 account (was a very long time ago). I recreated or reactivated the account and my files were still there and i got instantly charged all thr fees backwards to the day i closed the account.

They reimbursed though the charge.

While not directly related to the issue, I had a question (and semi-rant) in my mind for a very long time:

Haven't this scale of federation of services, "move fast and break things" approach, first to market races and rise of ready to use components and related machinery (containers, k8s, AWS, et al.) abused some well defined norms about technology and as a result aren't we in a somewhat broken state?

In the older days, we'd never have this kind of thing overlooked. You just weren't able to. Now things are much more complicated and this complicated services are assembled in a haphazard way to form big applications and platforms (e.g. too lazy to bind a folder with proper config to enable SSL, so put an NGINX https bridge container in front of it). As a result, the edge cases are not as edge anymore and breaking the configuration of a webapp is much easier.

Similarly this onion structure of abstraction over abstraction makes debugging and mending things much harder. To fix a one-off bug in a DB operation, you need to go to relevant container inside the pod, find its DB location, access if you can and update the table. That table maybe anywhere on this planet. This is impossible for a support engineer in most cases since we have convoluted SSO configurations which makes this kind of access impossible and the original developers couldn't care less in many cases.

I'll sound old, but even the new capabilities are fascinating, we have gone backwards in a lot of things, software quality and robustness related IMHO.

Couldn’t agree more with this sentiment and will also add that it happens on other aspects of life such as services.

The other day I called a benefits company that was supposed to send me a card but didn’t.

When I called they said there was an issue with their system which they would request to be fixed and that I should call in a couple of days to talk about the original request.

That happened after almost 2 hours waiting on the phone to be able to get to someone.

Before we didn’t have all these cool services but companies would do their whole jobs and get back to you when something goes wrong. Now, due to all these complexity and workers not keeping up with it things go way worse for the customer when something breaks.

I had the exact same thing happen to me, AWS did delete half of the charges, but still left me with 700$ on the account. So obviously, now AWS isn't my first go-to place to do business.

Do you have any recommendations for who's better? Isn't Google's customer service supposed to be pretty bad too?

I did gcp tutorial that left some resources running for a month. I eventually got charged around £300, I was able to talk to a human being and they sorted it out and refunded me. So they're still in my good books.

Awesome good to hear. Thanks.

I am pretty sure that they would lose in a small claims court, but probably most people who faced that can't bother

these conditions seem like an exploit in the rough