SBD: We Have a Filesystem at Home | Hacker News Reader

SBD: We Have a Filesystem at Home

fly.io

41 pointstorutofu29 comments

I really like fly.io posts, they're usually written in such a way that feels so human. I used to read all of them, because the writing was interesting, and secondarily sometimes the topic was.

Except seeing sentences like this:

> The difference is that a Sprite now stops and says so instead of carrying on with a disk it can’t vouch for

Makes it feel distinctly LLM written. Maybe I'm being RL'd to identify LLM authorship, maybe the author was RL'd to write like that by reading so many Claude-written works, but I'd like to see the humanity of fly.io blog posts come back.

I wish we'd stop getting replies like this to articles. It's just as tiring as LLM written articles.

People seem to forget that LLMs copied human writing. They also tended to copy people with more proper or formally trained writing styles. There are so many journalists and writers that talk about how they've had to 'stupify' how they write so they don't get accused of being an LLM.

It's like we're running idiocracy on fast forward.

Why? Why should we give up being critical of things we read?

Because being critical in those instances is more like going to mcdonalds and complaining about burger not being cooked to your specs. No one says you can't, but it does get tiring.

We expect McDonalds to be consistently mediocre because they do a lot of work to achieve this. Why should we expect every blog post on the internet to be consistently mediocre?

Um.. have you looked at most blogs? A vast, vast majority is not merely mediocre. I am not even exaggerating. Pareto principle applies. Post AI, triply so.

Sure, the vast majority of blog posts these days are AI-generated SEO slop. But this site isn't supposed to be a random sample of the web. The idea is that people upvote quality, but that doesn't seem to be happening. I don't know. Maybe HN is over and I should just give up on it.

It is a reasonable objection. I think I agree with that one.

Like going to an "independent" restaurant and being served a McDonalds Spicy Chicken Crispy Deluxe . Good sandwich, but I could have just gone to McDonalds.

And if McDonalds started mass-producing the restaurant's signature burger, I would hope the chef has enough pride/business sense to make a new signature burger.

>Why? Why should we give up being critical of things we read?

Remember how pre-AI everyone used to complain about hating how this or that website hijacked scrolling? Countless words were written about hijacked scrollbars and back buttons. Zero conversations were improved by the comments and zero websites were changed as a result. It was just complaining.

"Boo hoo AI" is the same kind of low-effort, complain about everything while adding nothing to the conversation post.

Instead, try to say something more substantive than "I found a word or phrase that I think was AI."

I wish I would have saved it, but there was something posted here a number of months back that people were saying had AI generated text and pictures (they could tell because of the pixels). Turns out it was an article from 2017.

It's just ironic to me how so many people are confidently incorrect about it, we are the LLMs in this case spouting 'facts' that we are sure of.

Was it this one maybe? https://news.ycombinator.com/item?id=48596910

Criticising something for sounding like an LLM is like criticising a gas-guzzler for being blue.

If it'd be any consolation, I for one find it interesting, even if you don't. I like to keep my AI writing detector honed, and others' opinions help. Even if I clicked figuring I wouldn't necessarily find the article all that interesting, I might end up reading it anyway just to see what I think of the writing style.

> I wish we'd stop getting replies like this to articles.

I wish we got more such replies like that to articles. This is a confusing times, and social interaction helps in several ways.

> People seem to forget that LLMs copied human writing.

This is literally not true, but more importantly the conclusion you trying to use it for is blatantly false.

Choosing how you write is a part of writing.

> It's just as tiring as LLM written articles.

It's not even as tiring as reading one unedited LLM article.

It is always worth remembering that LLMs use certain phrases and patterns because many source materials in the training set did. One or two smells does not a cesspool make.

As anti-AI as I am, people pointing out that they've spotted an AI are getting quite tiresome too. If you think people won't spot it without you telling them, then it is good enough for them no matter who or what generated it.

people are just going to have to turn the accursed spell-check off and start writing like psychos again. said modality was second nature online for like the first 15 years of the web. phones ruined us

Noticed it too. Feels like an LLM that ingested all old blog posts and did it's best to match the style. But it doesn't feel human anymore.

I’d cut them some slack if that’s the only example you found in the whole article. A lot of the “LLMisms” are actually valid english constructs* , it’s just that LLMs will use them in every sentence. I noticed while reading older stuff that’s way pre-llm, occasionally a sentence will jump out to me as LLM written, but it’ll be like once per chapter. When it’s every other sentence only then do I discount the entire thing as LLM written.

*: Except for the “Random label: sentence that has nothing to do with the label” construct. That one drives me crazy and it makes no sense

Reads like AI to me, or AI-shaped corposlop

> I have unkind things to say about people who make their own pasta

You lost me on the first sentence; my significant other's homemade ravioli is unparalleled. Homemade pasta is a delight to make and tastes so much better than $1 cardboard box pasta

Making your own pasta is easier and faster than baking your own bread.

I'll be honest, I don't understand what Sprites are, or why they need their own kind of block device. All of the pain seems to come back to trying to map a linux filesystem running on a block device into an object store. They could just ... not do that?

Here's why they do:

https://fly.io/blog/design-and-implementation/

But it feels like maybe they made the wrong tradeoff here.

A story which may or may not be based on lived experience: one day a Sprite stops booting. JuiceFS is asking object storage for a chunk and getting a 404. We wanted to go fast, so we ran JuiceFS in writeback mode. It records a chunk in its metadata first and uploads the bytes after. If the machine is killed at any time between those two steps, the metadata points at a chunk that doesn’t exist.

So we built a harness that drives fleets of Sprites through write, checkpoint, sleep, wake and restore, over and over, verifying every byte that comes back. We pointed it at 400 JuiceFS Sprites for about 5,000 cycles. A nonzero number of them permanently lost the ability to checkpoint. Well … fuck.

Enter: the Sprites Block Device (SBD). It takes a bucket and presents it to the Linux kernel as a block device, over NBD. The Sprite formats that device as ext4 and mounts it, the same way you’d mount a disk on any Linux box. Duh.

This blog has has a strange vibe like "the last BRAWNDO we made in a bathtub may or may not have killed some customers lmao, try NEW BRAWNDO 2.0 TODAY!" that I didn't realize could work for corporate messaging around an in-house CoW cloud filesystem.

While I love the post, I just kinda have to lol that it all ends up back where we started with ext4 atop a block device.

The more things change, the more they stay the same.

Now look, I’m not about to try and stop folks from finding new ways of solving problems, especially in infrastructure. Far from it, I love seeing what you beautiful diamonds come up with that works for you. They make for amazing reads!

Except 99.999% of infrastructure needs to be boring, bland, basic, and bulletproof. Most Infra folks will take the minor inefficiencies associated with object storage because we’re not a unicorn startup where solving such a problem could save us millions of dollars a year in write calls to our hypercloud provider; we’re far too busy trying to get our internal customers to stop overscheduling VMs or implement the latest executive shiny project. We need our underlying abstractions - hypervisors, storage arrays, network cores - to work, without fuss and with minimal maintenance.

So while it’s super neat that Fly.io has shared this, and they absolutely have a relevant and beneficial use case for reinventing the proverbial wheel, it’s important to note they are the edge case, not the norm.

I wonder if the folks at fly.io are aware of ublk (https://docs.kernel.org/block/ublk.html). It seems a perfect fit for this use case - I imagine their scale would benefit from a uring-based in-kernel driver that supports zero-copy thru the dataplane.

So it's like Archil or ZeroFS but more proprietary and less useful. Got it.

Look, we get it: anyone can write a filesystem now. It doesn't mean you should.

How is SBD like any of those?

Reply on news.ycombinator.com