Announcement

Collapse
No announcement yet.

systemd 254-rc1 Deprecates SysV Scripts, Adds Soft Reboots & systemd-battery-check

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • systemd 254-rc1 Deprecates SysV Scripts, Adds Soft Reboots & systemd-battery-check

    Phoronix: systemd 254-rc1 Deprecates SysV Scripts, Adds Soft Reboots & systemd-battery-check

    Out today is the first release candidate of systemd 254. This systemd update is another big feature release with new settings, new components, and many additions to existing systemd components. Plus there is some deprecations and breakage ahead for future releases...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    I see there's at least developers from: Microsoft, Meta and IBM/Red Hat.

    There are many developers from those three.

    What's into systemd? Despite people hating systemd, I found it interesting.

    First Lennart Pöttering joins Microsoft. Then Red Hat attitude gets even worse and now after reading the git log make me think that maybe depending on very strongly corporate backed projects is a really bad idea.

    I'm not saying systemd is bad per se, it has good stuff and other bad ones at the technical level. But strategically makes me wonder if depending on it is a good idea.

    I dislike those three corps. I don't believe Red Hat/IBM is good for the FOSS ecosystem too, for example. I don't trust Microsoft all all. And I don't trust Meta too.

    What do you think? Please research what the contributors and sponsors are in systemd, you might get really surprised.
    Last edited by timofonic; 07 July 2023, 11:27 AM.

    Comment


    • #3
      For anyone using VMware VM guests, especially on KDE where `systemd-xdg-autostart-generator` is used for XDG `.desktop` autostart files being converted into systemd units, a fix / workaround landed for systemd v254 that removes the timeout for starting autostart services.

      Probably a bug with the VMWare executable launched, but it's rather inconsistent with being detected as started by systemd, which could trigger the timeout and kill the process even though it was actually working perfectly fine. Seemed to be more likely when the CPU was under heavy load or the binary executed too early as delaying the start by a few seconds would reliably work, something racey going on.

      Been a known issue for KDE since Plasma 5.25 I think when they defaulted to systemd startup units. Presumably Gnome will adopt that too at some point.

      ---

      I liked how the systemd devs were quite responsive to resolving that when I reported the issue. Whereas with all the info I provided VMWare at the `open-vm-tools` repo, I've heard nothing back still.

      Comment


      • #4
        soft-reboot seems like a nice foundation to implement system updates on shutdown/reboot without needing to reboot the whole system. Right now Fedora reboots on a different target, which is not ideal, specially if you are using FDE as it prompts for your password again just for the updating screen.

        Comment


        • #5
          systemd-battery-check
          Having just looked at a system with a bad battery on a RAID adapter it makes me wonder if this could be extended with additional functionality/triggers for other targets than just a 'system laptop battery'.

          I personally like systemd and have used it extensively in VMs, bare metal, laptops, containers (docker/podman both), and I get that other people have had some bugs with it that I haven't had. And Poettering is not exactly friendly or nice.

          For me, It was hard to learn to use journalctl and how the unit files are shipped and stored and linked over and how to write my own. And even worse in the early days before people Got Used To It, debugging someone elses copy-n-pasted unit file they got from a stackoverflow answer.

          But if you ever had to debug even WORSE init.d scripts and rearrange ordering on a half-dozen services (and like me didn't experience these catastrophic omg things others complain about) you'd be thankful your fellow corpo slaves no longer had to torture you with their 25+ yr old init scripts from hell passed on down the years and modified from one tomcat project then to another that was similar 'in spirit', etc. (I didn't do it, and no longer work there). The systemctl commands are now the same everywhere, i no longer need to remember what the init commands were on debian vs chkconfig or whatever in old RHEL. I know the ordering for units will Just Work (normally)... i know how to find logs based on the unit name (normally). no more wondering where logs are.

          I get people don't like it, not trying to force ANYONE to like it, but 9 out of 10 distro parents agree, linux with VitaminSystemD is great for computing.

          Comment


          • #6
            Originally posted by timofonic View Post
            I see there's ay ñeast developers from: Microsoft, Meta and IBM/Red Hat.

            <snip>

            What do you think? Please research what the contributors and sponsors are in systemd, you might get really surprised.
            So I think what you're saying is you're surprised the people who feel the pain of init systems at their grandest and largest scale wanted to improve it? How many orgs actually pay people to work on *foundational* open source projects like this? I mean sure you'll find web libraries and occassional debug tools from corpos but this level requires Big Linux Users. Most of Home Devs <me anyway and most i know> aren't working on Kernel/Init/Boot/EFI/drivers (most). Its corporations.

            Anyway, substitute what you said about the linux kernel.
            Please research what the contributors and sponsors are in the linux kernel or grub or your favorite kernel module, you might get really surprised.
            Its corporate turtles all the way down until you get to the foundations, its more about ensuring your *way of computing* remains free, not who wrote it.

            Comment


            • #7
              Originally posted by timofonic View Post
              I see there's ay ñeast developers from: Microsoft, Meta and IBM/Red Hat.

              There are many developers from those three.

              What's into systemd? Despite people hating systemd, I found it interesting.

              First Lennart Pöttering joins Microsoft. Then Red Hat attitude gets even worse and now after reading the git log make me think that maybe depending on very strongly corporate backed projects is a really bad idea.

              I'm not saying systemd us bad per se, it has good stuff and other bad ones at the technical level. But strategically makes me wonder if depending on it is a good idea.

              I dislike those three corps. I don't believe Red Hat/IBM is good for the FOSS ecosystem too, for example. I don't trust Microsoft all all. And I don't trust Meta too.

              What do you think? Please research what the contributors and sponsors are in systemd, you might get really surprised.
              So do you mean that corporates should develop proprietary software instead of contributing to open source? Free software has always been about your freedom as a user. It has never been intended as some anti corporate protest.

              Comment


              • #8
                Originally posted by timofonic View Post
                I see there's ay ñeast developers from: Microsoft, Meta and IBM/Red Hat.

                There are many developers from those three.

                What's into systemd? Despite people hating systemd, I found it interesting.

                First Lennart Pöttering joins Microsoft. Then Red Hat attitude gets even worse and now after reading the git log make me think that maybe depending on very strongly corporate backed projects is a really bad idea.

                I'm not saying systemd us bad per se, it has good stuff and other bad ones at the technical level. But strategically makes me wonder if depending on it is a good idea.

                I dislike those three corps. I don't believe Red Hat/IBM is good for the FOSS ecosystem too, for example. I don't trust Microsoft all all. And I don't trust Meta too.

                What do you think? Please research what the contributors and sponsors are in systemd, you might get really surprised.
                with every new release of systemd the list of features that will never be used by 99.9% of machines grows. systemd is racing to it's inevitable replacement. This isn't a normal software trajectory. On the other hand they are replacing hodgepodge of shell scripts and various minor utilities into a cohesive system, so it has been a worthwhile effort. I think if and when systemd unifies network configuration and system booting, then people will start to look back and say, okay, but can we do this in a simpler way?

                Comment


                • #9
                  Nice features, by-this-way systemd is going to dominate over linux world and be a must dependency for it, i don't know it's good or bad.

                  Comment


                  • #10
                    Originally posted by fitzie View Post
                    with every new release of systemd the list of features that will never be used by 99.9% of machines grows. systemd is racing to it's inevitable replacement.
                    Whatever, man.
                    Wake me up when the replacement is delivered and relevant.

                    Comment

                    Working...
                    X