• As an admin, I prefer no swap on prod machines because I’d rather have the oom killer kill a process that will automatically be brought back up or replaced than grind everything to a crawl swapping. A dead process can be restarted. A swapped to death server can be challenging to even get into.

    • Onihikage@piefed.social
      link
      fedilink
      English
      arrow-up
      17
      arrow-down
      1
      ·
      11 days ago

      EarlyOOM is great for keeping systems responsive. I can’t understand why the default memory management on many distros still seems to be “do everything possible to avoid automatic termination of processes even if that means the system becomes borderline unusable.” It makes for a terrible user experience, and most users are just going to restart the machine when it happens rather than try to struggle through a slide show to manually kill whatever’s causing the problem.

      • backgroundcow@lemmy.world
        link
        fedilink
        arrow-up
        1
        arrow-down
        1
        ·
        9 days ago

        I couldn’t agree more. If there only was a somewhat user-friendly setting that allowed the oom killer to be far more aggressive, killing or freezing processes as soon as their memory use starts to affect system responsiveness, and just tell me this is what has happened.