Announcement

Collapse
No announcement yet.

Canonical Releases Upstart 1.6

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

  • #31
    Originally posted by ryao View Post
    I doubt anyone here will listen to me, but what Canonical is doing in Ubuntu is actually better for system reliability. Developers capable of reviewing systemd on technical merits find that it makes Linux systems less reliable. I do not think that the systemd developers even attempt to dispute this. The issue is that the kernel will panic if PID 1 dies and systemd's design makes it very difficult to avoid failures in PID 1.
    Theoretically less reliable or there were people that had actual problems with it?? Because from what i read around its pretty much rock solid.

    Comment


    • #32
      Originally posted by ryao View Post
      I doubt anyone here will listen to me, but what Canonical is doing in Ubuntu is actually better for system reliability. Developers capable of reviewing systemd on technical merits find that it makes Linux systems less reliable. I do not think that the systemd developers even attempt to dispute this. The issue is that the kernel will panic if PID 1 dies and systemd's design makes it very difficult to avoid failures in PID 1.
      Um, where do you get that from? Is this some kind of theoretical argument? Because it certainly isn't reflected in reality... systemd is out there being used by distros today, and it's proven as solid as any other init daemon.

      Comment


      • #33
        Originally posted by Delgarde View Post
        Um, where do you get that from? Is this some kind of theoretical argument? Because it certainly isn't reflected in reality... systemd is out there being used by distros today, and it's proven as solid as any other init daemon.
        I get it from being a distribution developer. It is what people that develop distributions that plan to avoid systemd have concluded.

        Comment


        • #34
          Originally posted by ryao View Post
          I get it from being a distribution developer. It is what people that develop distributions that plan to avoid systemd have concluded.
          Ok, so you're one of these "developers capable of reviewing systemd on technical merits"? Then can you explain your comment further - tell us what you actually mean when you talk about the fundamental flaws in it's design? Because I'm very surprised by your claim to know of significant flaws that even the systemd developers would acknowledge.

          Comment


          • #35
            Originally posted by 89c51 View Post
            Theoretically less reliable or there were people that had actual problems with it?? Because from what i read around its pretty much rock solid.
            Take a look at archlinux forums, and you'll find out how much "rock solid" it is.

            Comment


            • #36

              Comment


              • #37
                Originally posted by Delgarde View Post
                Ok, so you're one of these "developers capable of reviewing systemd on technical merits"? Then can you explain your comment further - tell us what you actually mean when you talk about the fundamental flaws in it's design? Because I'm very surprised by your claim to know of significant flaws that even the systemd developers would acknowledge.
                I bet he mean, every one should use openrc. The gentoo camp didn't like the inclusion of udev in systemd and the consolkit replacement. I got the impression that Lennart is kinda unpopular on the gentoo mailing lists...

                But as I understand the "Gnome Os" stuff. The idea is to homogenize the base system and have harder requirement on it. I suppose systemd is a potential start on this.
                Last edited by Akka; 11-16-2012, 01:53 AM.

                Comment


                • #38
                  Originally posted by Delgarde View Post
                  Ok, so you're one of these "developers capable of reviewing systemd on technical merits"? Then can you explain your comment further - tell us what you actually mean when you talk about the fundamental flaws in it's design? Because I'm very surprised by your claim to know of significant flaws that even the systemd developers would acknowledge.
                  The more complex that you make software, the more likely it is to fail. I would be surprised if the systemd developers claimed that systemd was as reliable as sysvinit. The chance of a bug killing systemd will be greater and being PID 1, that will kill the system.

                  Comment


                  • #39
                    Originally posted by ryao View Post
                    The more complex that you make software, the more likely it is to fail. I would be surprised if the systemd developers claimed that systemd was as reliable as sysvinit. The chance of a bug killing systemd will be greater and being PID 1, that will kill the system.
                    Solution: have the init system fork, exec systemd on pid 2, and on pid 1 close stdin, open /dev/null, and exec /bin/true. UNCRASHABLE LINUX LOLOLOLOL~!~!!

                    Comment


                    • #40
                      Originally posted by a2r-l View Post
                      Take a look at archlinux forums, and you'll find out how much "rock solid" it is.
                      Configuration problems =! unstable software.

                      Comment


                      • #41
                        Originally posted by snadrus View Post
                        SystemD (although having more purposes than Upstart) is nearer to the UNIX philosophy: "Do 1 thing & do it well" by ending the micromanagement once required of boot systems Upstart and Sys-V
                        systemd is everything but the kitchen sink. How is this suposed to be "Do 1 thing & do it well"?

                        Comment


                        • #42
                          Originally posted by elanthis View Post
                          Solution: have the init system fork, exec systemd on pid 2, and on pid 1 close stdin, open /dev/null, and exec /bin/true. UNCRASHABLE LINUX LOLOLOLOL~!~!!
                          If you loop fork, exec of systemd, it won't panic, but all systemd's child processes would also die.

                          Comment


                          • #43
                            Originally posted by LightBit View Post
                            systemd is everything but the kitchen sink. How is this suposed to be "Do 1 thing & do it well"?
                            Because it's a modular set of tools. There's systemd-journald, systemd-udevd, systemd-logind, systemd-readahead and various other components like systemd-fsck, systemd-localed, systemd-hostnamed, systemd-timedated among other things. Each do one thing and one thing well. Only the core part of systemd is runs at PID1 and the others behave and are launched like every other service or a tool.

                            Comment


                            • #44
                              Originally posted by Teho View Post
                              Because it's a modular set of tools. There's systemd-journald, systemd-udevd, systemd-logind, systemd-readahead and various other components like systemd-fsck, systemd-localed, systemd-hostnamed, systemd-timedated among other things. Each do one thing and one thing well. Only the core part of systemd is runs at PID1 and the others behave and are launched like every other service or a tool.
                              Even the core does more things.

                              http://cgit.freedesktop.org/systemd/.../tree/src/core
                              vs
                              http://svn.savannah.nongnu.org/viewv...?root=sysvinit
                              vs
                              http://www.openbsd.org/cgi-bin/cvsweb/src/sbin/init/
                              vs
                              http://plan9.bell-labs.com/sources/p...src/cmd/init.c

                              Comment


                              • #45
                                Originally posted by LightBit View Post
                                Even the core does more things.
                                Sure. You don't get the socket activated and heavily parallelized boot process and other similar modern features for free. Still if I'm not mistaken the PID1 part is (mostly) responssible for just launching daemons in various different ways.

                                Comment

                                Working...
                                X