this post was submitted on 30 Aug 2024
24 points (90.0% liked)

Selfhosted

40400 readers
791 users here now

A place to share alternatives to popular online services that can be self-hosted without giving up privacy or locking you into a service you don't control.

Rules:

  1. Be civil: we're here to support and learn from one another. Insults won't be tolerated. Flame wars are frowned upon.

  2. No spam posting.

  3. Posts have to be centered around self-hosting. There are other communities for discussing hardware or home computing. If it's not obvious why your post topic revolves around selfhosting, please include details to make it clear.

  4. Don't duplicate the full text of your blog or github here. Just post the link for folks to click.

  5. Submission headline should match the article title (don’t cherry-pick information from the title to fit your agenda).

  6. No trolling.

Resources:

Any issues on the community? Report it using the report flag.

Questions? DM the mods!

founded 2 years ago
MODERATORS
ruud
Loki
CannaVet
devve
HybridSarcasm
[email protected]
24
Silverblue or other immutable on remote VPS? (self.selfhosted)
submitted 2 months ago by aordogvan to c/selfhosted
29 comments fedilink hide all child comments
 

Does anyone know of a hosting service that offers Silverblue as a possible choice for OS?

It seems to me that for a server running only docker services the greatly reduced attack surface of an immutable distro presents a definitive advantage.

you are viewing a single comment's thread
view the rest of the comments
[–] [email protected] 9 points 2 months ago (2 children)

Because even if an attacker could gain access even as root he cannot modify system files.

They 100% can.

[–] [email protected] 1 points 2 months ago

Absitively, use case here IMO is set and forget autoupdate to stay current and SELinux (which actually reduces surface)

[–] asap 0 points 2 months ago* (last edited 2 months ago) (1 children)

They 100% can.

An attacker escaping from a container can't be system root as Podman runs rootless (without some other exploit or weak password).

The filesystem itself is also read-only.

/dev/nvme0n1p4 on /sysroot type xfs (ro)
/dev/nvme0n1p4 on /usr type xfs (ro)
/dev/nvme0n1p3 on /boot type ext4 (ro)
[–] [email protected] 5 points 2 months ago* (last edited 2 months ago) (1 children)

An attacker escaping from a container can’t be system root as Podman runs rootless (without some other exploit or weak password).

That would be true of podman running anywhere, and is not unique to an immutable distribution.

The filesystem itself is also read-only.

You can change that real quick if you have root access.

[–] asap 1 points 2 months ago* (last edited 2 months ago) (2 children)

edit: "Immutable" means "all of them are the same", not "unchangeable".

~~You sound confident, but the fact that Fedora is using the term "immutable" makes me wonder if you actually have domain expertise here.~~

~~Immutable means immutable. It would be strange for them to call it that if it actually means "completely irrelevant from a security perspective".~~

~~Unless you provide some evidence to the contrary I'm going to assume you aren't correct.~~

[–] [email protected] 6 points 2 months ago (1 children)

The immutability isn't designed to protect against a malicious attacker with root access.
Any system is fucked if that happens.
It's designed to reduce the workload of the maintainers, because they effectively only need to test and build for one standard image.

[–] asap 3 points 2 months ago

Makes sense. An "immutable" distro provides no additional security benefit, however CoreOS does have a reduced attack surface area compared to other distros, which itself is a benefit.

[–] [email protected] 2 points 2 months ago (2 children)

Someone with root can run ostree admin unlock --hotfix to make /usr writable. Someone with root can also delete all restore points.

It would be strange for them to call it that if it actually means “completely irrelevant from a security perspective”.

See the comment by superkret.

[–] aordogvan 1 points 2 months ago* (last edited 2 months ago) (1 children)

While what you're saying is theoretically true, don't forget that as far as I know, most attacks are perpetrated by bots. And while it is true that in a fedora based version one could run ostree admin unlock etc... this particular command would need to be included in the attack script.

Now if the script has to be modified to include all possible different immutable systems that could possibly run it would increase the complexity and most importantly the size of said script making it easier to detect.

I'm not saying that its a bulletproof method, I'm just saying that by itself it greatly minimizes the risk, at least until all servers run immutable systems. And even then it still complicates matters for potential attackers quite a bit. So therefore reducing or at least greatly minimizing the potential of the system being compromised.

[–] [email protected] 1 points 2 months ago

Because even if an attacker could gain access even as root he cannot modify system files.

Your comment was already from the position of if an attacker could gain root access. My responses were to that directly, and nothing else.

[–] asap 1 points 2 months ago (1 children)

While you are correct, any system is compromised if you have root, so isn't that irrelevant at that point?

[–] [email protected] 1 points 2 months ago* (last edited 2 months ago) (1 children)

While you are correct, any system is compromised if you have root, so isn’t that irrelevant at that point?

The original context for the comment chain was:

Because even if an attacker could gain access even as root he cannot modify system files.

So no, it's completely relevant.

[–] asap 1 points 2 months ago (1 children)

My comment in the comment chain was:

An attacker escaping from a container can't be system root as Podman runs rootless (without some other exploit or weak password).

We could give the op the benefit of the doubt and thinking that they were saying that the attacker inside the container managed to gain root inside the container.

[–] [email protected] 2 points 2 months ago (1 children)

Your comment also contained

The filesystem itself is also read-only.

Which is what led to the further discussion of root making that not so.

I don't believe that to be the intent of the OP's comment, given their second sentence, but they are welcome to state otherwise. I just don't want them thinking that an immutable distribution gives them some kind of bulletproof security that it doesn't.

[–] asap 1 points 2 months ago

Very true. The discussion helped me, as I did think it meant not easily editable.

As root of course you can change the system to be any other type of system (layer packages, rebase, whatever), but I did assume it meant not easily modifiable in it's current state.