Announcement

Collapse
No announcement yet.

systemd Rolling Out "run0" As sudo Alternative

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

  • #81
    Originally posted by access View Post

    Heh, that's the most powerful feature of systemd-resolved for me that you can have per interface DNS-servers/domain and get sane lookups when remoting in to e.g. work. I currently have 3 active VPN connections with separate domains and DNS servers and everything works beautifully together with my local DNS server.
    That's interesting. My friend hooked up a wireguard VPN to remote into work and that worked fine, but no matter what he did, resolved refused to do any lookups for traffic that was sent over the regular interface. I just assumed it was a security feature where it doesn't send data over normal interfaces if it detects a VPN connected. A quick switch to another DNS resolver fixed it instantly, with no configuration necessary.

    Comment


    • #82
      Originally posted by cynic View Post

      wrt the "DNS bug" is not a lack of man power keeping it open, it's not recognizing it as a bug.
      systemd dev think this is the way it is supposed to work.
      On the contrary, the entry is still open at Github and if one actually looks there it turns out that there is a problem reproducing the error and no user have so far added any debug logs despite it being constantly asked for.

      Comment


      • #83
        Originally posted by Kjell View Post
        systemd is vendor lock-in at this point

        You can't opt-out of most features unless you compile it yourself

        Packages are gradually starting to depend on functionality of systemd

        We no longer have the freedom of choice as they're pushing more and more features into a single point of failure.. XZ vulnerability depended on functionality of systemd (edit: for those missing the bigger point, such problems are a biproduct of feature creep. Also, let's not discuss how buggy core components are, like systemd-networkd & systemd-resolved, and how many attack vectors systemd introduces with the amount of modules they provide).

        How far will this go?
        systemd/GNU/Linux
        ​
        FreeBSD doesn't use systemd and os truly free and simply wonderful, why don't you switch to it?

        Comment


        • #84
          Originally posted by KernelCrasher View Post

          FreeBSD doesn't use systemd and os truly free and simply wonderful, why don't you switch to it?
          Because it doesn't use systemd and instead relies on shell script garbage.

          Comment


          • #85
            Originally posted by F.Ultra View Post

            On the contrary, the entry is still open at Github and if one actually looks there it turns out that there is a problem reproducing the error and no user have so far added any debug logs despite it being constantly asked for.
            The endless Github issue discussion I am referring to is (IIRC) about the dns query policy that is different from what all the rest of the world is doing.

            With glibc (and in all operating systems), each query was sent to all DNS servers, in the order they are specified, until one of them can resolve it. With systemd-resolved, the same server is always used until it fails. In that case, systemd-resolved will start submitting the queries starting from the second one, ignoring the first."

            This different policy causes unexpected behaviour in complex networks.

            So, AFAICT, there no real bug in the code.

            Comment


            • #86
              Originally posted by jacob View Post

              Because it doesn't use systemd and instead relies on shell script garbage.
              Tell that to Netflix, they surely forgot to consider such insightful and important point before choosing the OS.

              Comment


              • #87
                Originally posted by NotMine999 View Post

                So by that logic those users should have NEVER LEFT the Windows and Apple ecosystems and polluted the Wonderful World of Linux where FREE CHOICE reigned in pre-systemdeath days.

                To distill the comments and diatribes that I find in this thread it comes down to this:

                system(death) obviously and clearly breaks the fundamental rule of UNIX where many small programs work together, each specializing in their own unique tasks, to solve problems much greater than themselves. UNIX-style users could clearly swap out different small programs with other similar programs, albeit with different names (but a symlink can solve that), and for whatever reason they wanted; that's the essence of FREE CHOICE in UNIX and Linux.

                Yes, users can still compile and build what they want out of system(death) and a few posters have clearly stated that certain distros tightly integrate system(death) into many aspects of that distro. At least Gentoo took a more rational approach when it came to system(death) that still allows a large amount of user choice, for example (borrowing from a poster here) systemd-boot or grub or whatever. system(death) would be acceptable to many if it and it's sycophant distros functioned on a level that allows FREE CHOICE, but they do not; they cater the the "I don't wanna Windoze or Fruit Cult crowd but I needs me a 'puter so I'z can play mez games on kewl Linux."

                Where traditional UNIX-oriented distros have lost "mindshare" is in failing to improve integration among non-systemd components so they work well together. Distros claim "lack of developer hours", "lack of maintainer hours", "lack of funding" or whatever while pushing it all back to the priginal program developers (who have similar challenges). That is the "opening" that the systemd leaders spotted, how they wedged themselves into the Linux World ... and now like the roaches that they are we will never be rid of the system(death) infestation.
                Since you have so much time, compile your own operating system with a non-systemd init system.

                Comment


                • #88
                  Originally posted by KernelCrasher View Post

                  Tell that to Netflix, they surely forgot to consider such insightful and important point before choosing the OS.
                  They didn't forget it. All of their streaming and application servers run on Linux on AWS which itself runs on Linux. It's the OCAs (ie content caching proxies) that run FreeBSD.

                  Comment


                  • #89
                    Originally posted by andyprough View Post

                    You don't want to do everything sudo does, that's the beauty of doas, you avoid many of the security issues. You are showing your lack of basic knowledge again.
                    That's only your opinion. Thankfully nobody asked you.

                    Comment


                    • #90
                      Originally posted by L_A_G View Post

                      No. This is just patently false and the back and forth hard dependencies between systemD packages have been well documented for years at this point. You not personally running into this issue is textbook "runs fine on my machine" denialism and ignorant at best, flat out dishonest at worst.

                      I'll give you the benefit of the doubt and assume you didn't know and do you a favor and save you the effort of looking this up.

                      There. Now you know better.
                      You gotta be trolling.

                      You're saying that "systemD is not modular, because other projects depend on it"
                      Is that systemD's fault? What the fuck.

                      Edit: https://dictionary.cambridge.org/dic...nglish/modular

                      Comment

                      Working...
                      X