• 0 Posts
  • 33 Comments
Joined 1 month ago
cake
Cake day: October 3rd, 2026

help-circle

  • Guiorgy@lemmy.mltoLinux@lemmy.ml•systemd really, really sucks
    link
    fedilink
    arrow-up
    4
    arrow-down
    1
    ·
    9 days ago

    The crux of the issue, at least as I understand it, is that traditionally you’d build the system components in a way that they were interchangeable by defining some standard-ish API. While you can interchange some Systemd components, since “ststemd-xyz” is just the umbrella name of the project as a whole, so some components are completely independent, the core ones can’t really easily be swapped, and it’s only getting worse, especially as systemd is starting to become a hard requirement, a lot of people are rightfully feeling peeved. There’s a great talk made by a BSD developer that admits the virtues of Systemd while bringing the downsides, so no blind hate, you can google for the video “The tragedy of systemd”.







  • None of my VM require sharing of data, so each gets their own. I don’t currently use ZFS in Proxmox, but do have snapshots and periodic backups to my NAS, so recovery is generally not an issue.

    Conceptually, mounting a host storage inside a VM requires the use of either virtiofsd or NFS daemon running on the host (depending whether you mount it as local or network storage), so if either of the two or the underlying filesystem has a vulnerability, it does in theory create an attack surface, which is not ideal. With that said, as I mentioned above, I have not research led this enough, since I haven’t had the need yet, so don’t quote me on that. At the end of the day, it’s all about where you personally draw the line, so if you are fine with that risk and absolutely can’t give up the feature than go for it, it’s your system.


  • If you asked me 2 years ago, I’d say they give a decent level of isolation for cheap, but, at the end of the day, in the eyes of the kernel, containerized processes are just another process running on the same host, just with a different set of privileges. If you have public facing services, or untrusted binaries, I suggest moving those to VMs, local services can stay in containers. This is what I did after Copy Fail and Dirty Frag were discovered. Alpine Linux VMs are pretty lean, so I went with that, and to keep the migration simple, I run the services as containers inside the VMs.


  • For home Wi-Fi I use a strong password, but that’s because it’s something I input once in a blue moon. For guests I have a guest Wi-Fi with a simpler* (still reasonably secure and random, but shorter and without special characters) password and a QR code for simple use, though they are rate limited and have no access to any other network clients.


  • If we take the EFF wordlist commonly used to generate passphrases, we have 7776 words. If we take ASCII alphanumerics and special characters we have about 92. Assuming we generate a NIST compliant password of 15 characters in length we get 92^15=2.8*10^29 of combinations. To get the same level of security (same number of combinations) using the EFF wordlist we need (log(2.8)+29)/log(7776)=29.45/3.9=7.6 around 8 words. If we assume the words are 7 characters long on average, plus the separators, we get 7*8+6=62 characters long. If we start to also include Unicode characters like other scripts or emojis (something I’ve recommended to others) the required number of words grows even more. I don’t know about you, but I can’t be bothered to type that all out every time, a password manager works for me.