moonpiedumplings

@ moonpiedumplings @programming.dev

Posts

28
Comments

541
Joined

3 yr. ago

11h ago

Setting up a testing lab
Jump
1

moonpiedumplings @programming.dev 11h ago
Do you use the web ui?
I use the web ui heavily, but it's only packaged by the incus package from the author, and not included in the debian packages.
Also, what are you using for authentication?

16h ago

Setting up a testing lab

moonpiedumplings @programming.dev 16h ago

I (plus friends who do something similar) have been using centralized auth systems for this stuff. Proxmox supports OIDC, so if you are using Authentik or something similar you can just use one password.

And then Authentik supports 2FA, so you can use TOTP with that, or use passwords only.

17h ago

Setting up a testing lab

Jump

moonpiedumplings @programming.dev 17h ago

In addition to netbox, a wiki or other knowledgebase would be nice. You can document setup procedures as you go, and then other people can use that to figure stuff out.

17h ago

Working to Decentralize FedCM

Jump

moonpiedumplings @programming.dev 17h ago

Forgejo has a feature (that people usually disable) where you can bring your own openid connect url and use it to auth. So if I have my own OIDC provider I am self hosting, I can just use that to log in.

Most people only use OIDC for google and microsoft and whatnot but it's very possible. I don't realkly see what FedCM offers that OIDC doesn't or can't, or why we shouldn't be adding features to the existing and popular OIDC instead.

21h ago

/e/OS is not a secure OS

Jump

moonpiedumplings @programming.dev 21h ago

The problem is that real dumb phones are hard to find. Many modern "dumb phones" are actually full android devices, complete with a boatload of spyware that helps keep the cost of the device itself low.

KaiOS is better but that's a whole linux distro, with similar issues.

Since you mentioned tethering, do you have an example of a non android (or at least one that's not preloaded with a ton of spyware) dumbphone that supports usb tethering? I am skeptical that a real dumbphone would have this feature.

22h ago

Ubuntu 26.04 LTS Officially Supporting Cloud-Based Authentication With Authd

Jump

moonpiedumplings @programming.dev 22h ago

My one fear with this is offline authentication. I enjoy oauth/oidc a lot, but it doesn't have mechanisms for machines to continue to be able to authenticate while offline, like the way ldap/kerberos can do.

Is this just for machines that will always be online? I can understand that usecase but :/

EDIT: Okay, one comment, mentions himmelblau an alternative to authd, which seems to be more mature. Himmelblau has docs about offline usage. It looks like it has an emergency config that can use a cached password from the oidc provider,

Single-factor authentication (SFA-only) users and Hello-PIN users already have offline sign-in capability

Hmmm. Okay. Upon doing further reseach, it looks like offline authentication is exclusive to Microsoft Entra ID. :/

22h ago

Note taking app that I can link between my laptop and phone ?

Jump

moonpiedumplings @programming.dev 22h ago

Syncthing has encryption as well. You can have a device be "untrusted" so you put in an encryption password, and data sent to and stored on that device will be encrypted.

Although this does encrypt file (and directory) names, the caveats about folder structure and modification time still apply.

1d ago

Is legal the same as legitimate: AI reimplementation and the erosion of copyleft

Jump

moonpiedumplings @programming.dev 1d ago

He fed only the API and the test suite to Claude and asked it to reimplement the library from scratch.

What was the test suite licenced under? If it was in the same repo, then it was probably LGPL code as well.

If the MIT rewrite uses the LGPL licensed test cases, including them in the repo, then it probably must be LGPL as well.

2d ago

GitHub - sergi0g/cup: 🥤Docker container updates made easy

Jump

moonpiedumplings @programming.dev 2d ago

I use fluxcd with helmrelease's which auto update the helm release. If the helm chart versions specify container versions, then updating the helm chart updates the containers in the deployments.

But for raw deployments, I found this, but not much else.

2d ago

Chance rule

Jump

moonpiedumplings @programming.dev 2d ago

https://en.wikipedia.org/wiki/Superghost_(game)

https://en.wikipedia.org/wiki/Ghost_(game)

3d ago

Keycloak or alternative?

Jump

moonpiedumplings @programming.dev 3d ago

In addition to adding more worker instances, you can also increase the amount of threads each worker instance uses to vertically scale. It's about equivalent to adding a worker instance.

3d ago

Keycloak or alternative?

Jump

moonpiedumplings @programming.dev 3d ago

Authentik is definitely the best of all I've tried. It has the most features, supporting both ldap and oauth, and also has an official helm chart.

6d ago

What sense is licensing operating system on BSD license?

Jump

moonpiedumplings @programming.dev 6d ago

Openbsd is definitely more secure than secureblue. There is only so much you can do to handle the massive monolithic architecture of the Linux kernel. Further down the stack, many parts of Linux, like sudo, dbus, or systemd are regularly hit by zero days. The SELinux domain architecture that Secureblue is interesting, but SELinux is extremely complex and difficult to get right, compared to the much more simpler pledge and unveil sandboxing that openbsd offers.

In addition to that, there are further issues like the problematic way that user namespaces interact with browsers. (And user namespaces are frustrating in general, secureblue actually has a short article on their problems). For maximum security, you want to sandbox tabs from eachother using user namespaces (only works on chromium btw, firefox can't do this so it doesn't matter) — BUT, if you run your browser in a sanbox created by user namespaces, then you can't nest them, disallowing you from using that powerful tool to isolate tabs. So you are forced to make a choice: You can either sandbox the browser itself, in exchange for weakening the isolation between tabs, or you can strengthen the isolation between tabs, in exchange for weaking the sandbox around the browser itself. Giving the browser access to user namespaces is questionable though, because see above, user namespaces have led to a lot of vulnerabilities.

OpenBSD's pledge + unveil (but only on chromium again), does not really make such tradeoffs. It can sandbox tabs from eachother, while also sandboxing the browser itself. In addition to that, pledge + unveil do not present a massive kernel attack surface that people have had to restrict for having too many 0days. And this is just one of the many, many examples, where OpenBSD presents a better security posture than Linux.

Qubes is technically Xen, a different kernel than Linux. The Xen kernel virtualizes Linux distros, from which you can manage Qubes/Xen, or do normal Linux app stuff. But nothing stops you from using a BSD virtualized by Xen for management or usage. Qubes talks about why they use Xen here — but the short version is that they did not consider the Linux kernel's kvm secure enough for their usecase.

6d ago

How to access home network (eg, VPN) without port forwarding?

Jump

moonpiedumplings @programming.dev 6d ago

Tailscale already does though, I think.

https://tailscale.com/docs/features/tailscale-funnel

Although it might work differently.

7d ago

Weird server requests

Jump

moonpiedumplings @programming.dev 7d ago

What port was this sent to, and what webserver are you running (if it was sent to a webserver)?

This reminds me of the string to strigger the really bad apache vulnerabilities that lead to being able to read from the whole filesystem (path traversal), or get a shell on your system (remote code execution). It's likelu that bots are spray and praying attempts across the internet. As long as you're up to date, you should be good.

1w ago

EA is hiring a Senior Anti-Cheat Engineer to lead development of a native ARM64 driver for their Javelin kernel anti-cheat system and start laying groundwork for Linux/Proton support

Jump

moonpiedumplings @programming.dev 1w ago

I've had similar experiences with this FPS game called krunker.io

Krunker.io is a browser based game, and it had a pretty bad cheating problem, and since it was a browser based gamr, the devs could never implement an anticheat that worked for long.

They implemented a deputization system, where certain respected members of the community would become "krunker police", and then you could call them from a lobby. They would then invisibly spectate, and record and ban cheaters. The system worked really well, actually. Cheaters were banned quickly, and the requirement collection of video evidence held those involved accountable.

But krunker players had another interesting way of handling cheaters. You see, krunker has really bad netcode, bad enough that you would have to lead hitscan weapons a variable amount depending on how much ping you had. Krunker was also a movement shooter, where you could slidehop and go really, really fasy. The combination meant that you could dodge the shots of cheaters. As I got better, I just stopped calling krunker police, and started beating them. One of my fondest memories was this one lobby full of good players, and when a cheater joined we stomped them below all of us on the ranking, taunting them all the way down. At the end, they tried to sell their cheats and we all laughed. "Why would I buy these cheats? I'm better than them". Eventually they ragequit. Good times.

But nooooo, nowadays modern game publishers need control over every part of the game. They demand control over the servers, refusing to let anybody host their own communities. They demand absolute control over the community, but refuse to actually moderate it and handle toxicity. And now, they're demanding control over the clients, forcing you to install rootkits on your computers so they can control those too.

1w ago

What sense is licensing operating system on BSD license?

Jump

moonpiedumplings @programming.dev 1w ago

FreeBSD, OpenBSD and NetBSD are behind Linux.

Look, I dislike permissive licenses too, but you need a source to back this claim up.

Right now, each BSD does something special, that Linux (distro's) can't trivially replace, even if the usecase is more niche. NetBSD Dev's make efforts to get it running on many devices as they can. OpenBSD (and it's subprojects) are highly secure, moreso than Linux. Who do you think makes our beloved OpenSSH? OpenSSH noted for having very few vulnerabilities over it's two decade long existence, and OpenBSD itself is similar, which is insane because there are products with multiple bad vulnerabilities every year (Linux being one of them...). This is due to a highly security minded architecture - one that Linux lacks.

FreeBSD is like Linux before systemd. I like systemd, but systemd is really trying to be kubernetes on a single node. I like systemd because I like kubernetes, but I understand why someone wouldn't like it, and I question if "single node k8s" is the best architecture for a single server or personal desktop. The ports system results in freebsd packaging many server services that aren't packaged on Linux. Being able to manage those through the system package manager, and the conviniences that provides, is nice.

Different, and not popular don't mean bad.

1w ago

Adding "Log In With Mastodon" to Auth0 - Terence Eden’s Blog

Jump

moonpiedumplings @programming.dev 1w ago

This reminds me of the way that forgejo lets you feed it an arbitrary openid url, so you can log in with any service you want, including your own server.

Also, is this compatible with lemmy? The last time I tried fediverse (mastodon) login, it was with owncast, but it didn't work with lemmy.

1w ago

How to Automate QEMU/KVM VM Deployments?

Jump

moonpiedumplings @programming.dev 1w ago

Have you used ovirt? It's currently being maintained by Oracle after Red Hat gave it up.

I've been meaning to try it, but the documentation is dense and hard to get through, and I unironically find the openstack install instructions more approachable in some ways...

1w ago

How to Automate QEMU/KVM VM Deployments?

Jump

moonpiedumplings @programming.dev 1w ago

My recommendation is to use an abstraction layer that runs qemu-kvm under the hood and automate that. Some people have mentioned libvirt, but Incus is another good option.