• 1 Post
  • 14 Comments
Joined 1 year ago
cake
Cake day: June 17th, 2023

help-circle

  • If you’re wanting to do something like that, you’re probably best running Proxmox as a bit of a hypervisor, then Yunohost in a Debian VM on top, and assign something like “home.domain.tld” to Yunohost and get your “stable” family services running.

    Then you can try out other stuff like Coop, Cosmos, OMV, Caprover, Tipi, etc as other VMs if you wanna try adding something Yunohost can’t or doesn’t do well. Or if you wanna extend your DevOps skills without messing up family-prod. I mean, you could even have another Yunohost as a “sandbox.domain.tld” before new service deploy.


  • I’ve had Yunohost running in some way for probably 4+ years? It’s relatively solid, I can mostly depend on it without any issues. I like the SSO/LDAP user auth and perms, and the default fail2ban and ability to change ssh port from the UI. The update and system services pages are nice.

    What I don’t like is how apps are all installed locally instead of using containers or VMs. And resources are shared, so if one app uses, for instance, MongoDB, and another app needs it as well, they have to share the same one. It makes things run a bit leaner, but I do worry a bit about data bleed if there’s some vulnerability. And the apps are really hit and miss, since they have to be packaged, managed, and issue-tracked independently for this platform instead of the main app/project. So you find lots of orphaned or half-maintained apps that should be great otherwise.

    So you either suck it up and deal, or become a bit of a hacker/maintainer yourself on apps you care the most about. But if I wanted to get that involved I’d just roll a manual build myself. I submit issues and try to help where I can, but that’s not where I want to be.

    You could probably install something like Portainer and manually edit the NGINX config/homepage to hack some docker in there, but idk if I care enough to do that.


  • Eh, it is what it is. I have a full family life and a job screwing with computers all week. I don’t want to deal with spinning up, troubleshooting, and maintaining a mini devops stack.

    I don’t want to spend so much personal time to keep up with all the management and config, but I don’t think that means someone like me should have to live in a big tech world. If there’s a good framework that helps keep things easy to manage and secure for a minimal amount of input and time, even if I could run most of it myself manually with a lot more time investment, there’s no reason not to, IMHO.


  • Yeah, I know they’re different. I was just giving some background about what was going on, sorry if I confused.

    Just wondering if anyone has used what seems to be their compose/swarm config tool “abra”, especially multiserver, and have any feedback about it. I like that it seems to be pretty agnostic after doing its work, they say you can backup and export the config and use it elsewhere mostly as-is. Just can’t see much anywhere else about it.






  • Most of this is right, but needs some things corrected.

    LOS is kept up by individual maintainers of the devices, and so it can cover more of them. But that also means you expand your attack surface to lineage, maintainer, microg, etc. And that’s just on supported devices. Unofficial devices are even more wild-west, having much delayed releases, OS updates, security updates, everything.

    Not only that, but Lineage requires that you unlock your bootloader and often have your phone rooted to be able to do everything. This introduces special points of insecurity and possible issues in the future.

    GOS is from a single source, for a single line of phones, and uses a designed method to load cryptographically signed ROMs onto the device, and then validate updates using the same method. The Play Services are sandboxed and disabled by default, so you can just never use them if you want. Overall, this makes for a more cohesive device. One that is more private and more secure. Especially so, when you can buy a new Pixel device and have guaranteed updates for as long as Google will do so for the same device.


  • I live at a place where I needed Starlink so I feel entitled to comment.

    Ordered, and it took 6-7mo to allow me to start. In the meantime T-Mobile Home Internet let me start immediately. I kept both because when one had issues the other would be better (storms, updates, tower maintenance, downtime, Russian attacks, etc). But I noticed that Starlink kept getting worse. Lower speed, worse jitter/ping/bufferbloat/etc. it would routinely fail to hit 100mbps down with good sky view, mounted to a pergola. TMHI would routinely be above 250mbps, and I move to using it more often. Eventually a local ISP got a grant to roll out FttH in my area and I got rid of both.

    It’s been a bit over a year since then, maybe things got better. But I noticed Starlink overselling their nodes, being non-communicative for support issues, and missing these easily attainable FCC goals to people that often have much less options than I did. There’s no reason for them to get absolutely wiped by a cell phone tower. Hope they made enough by packing on customers, because they just lost $900m





  • I remember Ars Technica had an article or series on his bad decisions called “Apotheker needs an Apothecary” and lit into him for all the dumb things he was saying and doing. I just don’t see how you can have the manufacturing and branding behemoth HP was then, get giftwrapped Palm and webOS while RIM was still in the process of imploding, and fumble the bag so hard