Re:An Analysis of GrapheneOS's Server Infrastructure
by mhitza on Sunday May 31, 2026 @12:01PM
(#48346783)
Raising good points about having hosted infrastructure in the US.
The article is a bit too verbose, with parts that really don't tell much. Almost as if llm written?
The argument that seasoned sysadmins install (Debian or) *Alpine* is way off. Alpine only became popular in container runtimes, and maybe die hard fans run it everywhere, but musl and limited packages is an inconvenience on server setups.
Re:An Analysis of GrapheneOS's Server Infrastructure
by jasonvorhe on Sunday May 31, 2026 @12:43PM
(#48347207)
Reads like the result of a one shot prompt to write an analysis based on one or many git repos.
Currently getting errors from the host write.as which seems weird:
> Sorry, but our Write.as servers are getting so many visitors that it's making us very, very sleepy. Please try again in a little while, once we've had some rest.
> Is this your post? We don't fall asleep as easily for Write.as subscribers.
Weird, they're expecting me to subscribe somewhere to be able read some random post?
Re:An Analysis of GrapheneOS's Server Infrastructure
by cmeacham98 on Sunday May 31, 2026 @01:07PM
(#48347399)
It actually astounds me this is getting any number of upvotes on HN at all - it's so obviously LLM written (or at minimum heavily edited) that it's written in markdown on a platform that doesn't appear to support markdown.
It literally hits almost every major identifier of AI writing. You could probably get similar output yourself with a prompt along the lines of "Write a blog post analyzing the GrapheneOS infrastructure based on public information such as GitHub repositories. Focus on contrasting the security of the phone OS with the unnecessary attack surface and similar decisions of the infrastructure repos."
Re:An Analysis of GrapheneOS's Server Infrastructure
by mschuster91 on Sunday May 31, 2026 @01:03PM
(#48347369)
> Every GrapheneOS server has a local DNS resolver (Unbound) that's configured not to resolve queries itself, but to forward them to Cloudflare's servers over an encrypted connection. This means Cloudflare sees the DNS query patterns of every GrapheneOS server — what domains they look up, when, and how often.
Well, the general idea is to protect the privacy of GrapheneOS users. Sure, a government-backed entity can wiretap the GrapheneOS servers and force Cloudflare to deliver DNS logs to then correlate with requests... but that is a class of attack that I really don't see anyone doing anytime soon.
The article is a bit too verbose, with parts that really don't tell much. Almost as if llm written?
The argument that seasoned sysadmins install (Debian or) *Alpine* is way off. Alpine only became popular in container runtimes, and maybe die hard fans run it everywhere, but musl and limited packages is an inconvenience on server setups.
Currently getting errors from the host write.as which seems weird:
> Sorry, but our Write.as servers are getting so many visitors that it's making us very, very sleepy. Please try again in a little while, once we've had some rest. > Is this your post? We don't fall asleep as easily for Write.as subscribers.
Weird, they're expecting me to subscribe somewhere to be able read some random post?
It literally hits almost every major identifier of AI writing. You could probably get similar output yourself with a prompt along the lines of "Write a blog post analyzing the GrapheneOS infrastructure based on public information such as GitHub repositories. Focus on contrasting the security of the phone OS with the unnecessary attack surface and similar decisions of the infrastructure repos."
Well, the general idea is to protect the privacy of GrapheneOS users. Sure, a government-backed entity can wiretap the GrapheneOS servers and force Cloudflare to deliver DNS logs to then correlate with requests... but that is a class of attack that I really don't see anyone doing anytime soon.