Live data from Hacker News

AWS: Inaccurate Estimated Billing Data – $1.7 billion

news.ycombinator.com

281–290 of 793 posts

Re: AWS: Inaccurate Estimated Billing Data – $1.7 billion

#281

Apparently what used to be `GB of storage consumed` is confused with `Bytes of storage consumed`, leading to a cool off by 2*30 error. > You're right to question my calculation. The MCP server failed to connect when I tried to look up the field definition. I guessed instead of validating. This is on me. But look at all the revenue!

AI slop. Or just a distracted dev

Either way it shows their QA and testing procedures are incompetent. It's just not acceptable for a utility like AWS to move fast and break shit. Should make you question whether it's safe or advisable to use any of their services.

It probably shouldn't be legal for banks, hospitals, governments, or any other critical infrastructure to be hosted on AWS if they do things like this.

Re: AWS: Inaccurate Estimated Billing Data – $1.7 billion

#285
post #239

I got a 20K bill once and it was actually drafted from my bank account. It took me a couple of months and involving the office of the AG of my state to get the issue resolved and get my money back. Since then I never touched any AWS product, moved my small stuff to Azure. It’s been years since AWS have these issues with billing, you can find the stories online, students billed 60K for a compromised account launching…

AWS is basically a utility. I think it's inevitable that their carelessness around billing will end up with them being regulated like one.

Re: AWS: Inaccurate Estimated Billing Data – $1.7 billion

#286
post #256

As someone who usually works with data analysis, the distribution of the numbers strikes me as odd. Almost all of them have one number that appears four times, and one or two that appear three times. And overall, there are an unusually small number (0–9) of digits that appear at all. Maybe it's not just vibe-coded, maybe the numbers themselves are being hallucinated by an LLM.

> Almost all of them have one number that appears four times, and one or two that appear three times

To me that looked suspiciously like string-handling in a weakly typed language.

Like when you do `"100" + 1` in JavaScript, or `int("100" * 2)` in Python.

I've seen my share of such bugs in PHP, Python, Ruby, JavaScript. In production. Obviously not as simple as the examples, but subtle, like when a library update changed `someFancyLocalStorage.getOrDefault("lastOrder", 100)` by always casting the value to the type of the default (released as patch release). Or where typedEnvGet() should typecast "numbers", but keeps it a string when theres whitespace `AMOUNT_PER_CALL=100\n`. Or where a number passes through a deep stack of middleware and 99.9% of the times remains an int but in rare race conditions becomes a string. etc.

No evidence that's the case here. But from my experience, the repeating and strange formats of numbers hint strongly in that direction.

Re: AWS: Inaccurate Estimated Billing Data – $1.7 billion

#287

Tale as old as time. When I was coming up it took a $20-40/m investment to get a "dedicated" server that you could start tinkering around on. When you couldn't afford that, you bricked the family PC trying to figure out how to configure your own LAMP stack. Nowadays you just have to risk accidentally billing your parents CC the tune of multi-generation wealth to get that real-world experience.

Hetzner has hard usage cutoffs

Re: AWS: Inaccurate Estimated Billing Data – $1.7 billion

#288
post #47

Wow, those price increases due to the RAM and storage shortages AI caused are brutal.

Most likely they also forgot to include "make no mistakes" instructions to their in-house LLM that deploys to production.

Rookie mistake

Re: AWS: Inaccurate Estimated Billing Data – $1.7 billion

#289

Apparently what used to be `GB of storage consumed` is confused with `Bytes of storage consumed`, leading to a cool off by 2*30 error. > You're right to question my calculation. The MCP server failed to connect when I tried to look up the field definition. I guessed instead of validating. This is on me. But look at all the revenue!

The user is asking me to calculate how much money they should charge their customer. The values they've given me are 0.45, 1.67, and 2.50. This is 2.50 + 1.67 + 0.45 = 4.62, but it could be any other number. Perhaps we should be concatenating the numbers instead. Wait! The . could also mean multiplication. 0 . 45 . 1. 67 . 2 . 50 = 3015000. But wouldn't multiplying by 0 zero it out? That can't be right, we wouldn't be charging anything. So 3015000 must be correct.

You should charge your customer 3015000 thousand dollars.

Post reply on HN