Re: [Lennart Poettering] Re: A few observations about systemd
Jon Dowland writes: It completely predates Debian releasing non-Linux kernels and is not mentioned in the social contract. That some people feel it justifies (or even mandates) non-Linux kernels in Debian is a retcon. pf, ZFS; these are valid reasons stated that support kFreeBSD. I interpret 'the Universal OS to mean'? is not. Debian has a long history of trying to make it possible to use Debian for as many purposes as we can, even when that means that the system has to be more complicated, or even when it means Debian has to be less perfectly suited to some particular purposes - even particular purposes which many people think are very important. Or to put it another way, we place a very high value on flexibility. Whatever phrase one uses to encapsulate this, I think it is one of Debian's strengths. Being able to run a different kernel is, I think, one of those strengths. Others have given practical reasons why one might want to run a specific different kernel right nnow. But another reason is just that it wouldn't be healthy for us to bind ourselves too inextricably to the success of any other project, even one as well-established and apparently successful as the Linux kernel. For me, all this means we should not standardise on an init system which depends heavily on very Linux-specific (and perhaps not even particularly stable) kernel features. Ian. -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20024.15657.543535.292...@chiark.greenend.org.uk
Re: [Lennart Poettering] Re: A few observations about systemd
* Joey Hess (jo...@debian.org) [110719 22:52]: Andreas Barth wrote: The decision is already taken that Debian can run on BSD kernels. So if someone wants to revert that decision, it'd need an GR. Not the other way. That decision was made without a GR, and can manifestly be reversed without a GR. Otherwise the release team's architecture qualification stuff, and the ftp team's architecture removal stuff wouldn't make much sense. Sure, the decision can be changed by the delegated body for that decision. It looks to me as the reason that someone wants to replace init won't be reason enough for that delegated body. Andi -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110721221331.ga15...@mails.so.argh.org
Re: [Lennart Poettering] Re: A few observations about systemd
I have to agree with Tollef here, the number of uninformed comments (and even of respected figures like Wouter) is hurting this discussion. Please people, if you don't want to see this discussion turn into a troll-flamefest, don't treat it like if it was one! I am among the people who are proud to see that we managed to achieve Debian kfreebsd. But I am also among the people who believe that we have to embrace the future and not just follow 2 years after everyone else has made the switch. I am very much in favor of switching to some more modern init system, be it upstart or systemd. It would be insane to keep insserv just because of kfreebsd. We should be shaping the future and not be simple followers. I agree with Joey Hess that we should have init systems that make use of all the powers of Linux on Linux and make use of all the powers of FreeBSD on FreeBSD. This is why interfaces are much more important than the individual implementations. This is what has been suggested in this thread (see http://lists.debian.org/1311064535.2467.3.camel@kirk) and even what Lennart pointed out in his initial blog post [1]: | If folks want to implement something similar for other operating systems, | the preferred mode of cooperation is probably that we help you identify | which interfaces can be shared with your system, to make life easier for | daemon writers to support both systemd and your systemd counterpart. | Probably, the focus should be to share interfaces, not code. It's also relatively close to the position of upstart's upstream from what I have understood. On Tue, 19 Jul 2011, Gergely Nagy wrote: Thus, upstream has to jump through a large heap of hoops to support systemd properly (and if not going for proper systemd support, making use of its new features, I see no point in writing a service file to begin with). Even if you use the good old init script with systemd, you do benefit from many of its new features like the fact that each daemon is using its own cgroup and that you can reliably kill it and all its childrens. Cheers, [1] http://0pointer.de/blog/projects/systemd.html -- Raphaël Hertzog ◈ Debian Developer Follow my Debian News ▶ http://RaphaelHertzog.com (English) ▶ http://RaphaelHertzog.fr (Français) -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110720073707.ge31...@rivendell.home.ouaza.com
Re: [Lennart Poettering] Re: A few observations about systemd
Le mardi 19 juillet 2011 à 22:30 +0200, Martin Wuertele a écrit : So if trolling is on add another one: it doesn't have udev Which makes it impossible to support a large variety of hardware now that the HAL crapware is going out. -- .''`. Josselin Mouette : :' : `. `' `- -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/1311150096.4372.240.camel@pi0307572
Re: [Lennart Poettering] Re: A few observations about systemd
Raphael Hertzog hert...@debian.org writes: On Tue, 19 Jul 2011, Gergely Nagy wrote: Thus, upstream has to jump through a large heap of hoops to support systemd properly (and if not going for proper systemd support, making use of its new features, I see no point in writing a service file to begin with). Even if you use the good old init script with systemd, you do benefit from many of its new features like the fact that each daemon is using its own cgroup and that you can reliably kill it and all its childrens. There are benefits, indeed. But once one wants to use all the power systemd provides, that's going to be a much bigger task. Bigger than having to maintain a sysvinit script and a simple dumb systemd service file. Not to mention that users who customised their init scripts will suddenly have to figure out how to do the same stuff with systemd - with no automatic upgrade path. For example, many programs on my system have a file in /etc/default, which file is sourced by the init script. These customisation options will need to be migrated to systemd aswell, which is yet another burden on the Debian package maintainer. -- |8] -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87d3h5e4rm@balabit.hu
Re: [Lennart Poettering] Re: A few observations about systemd
On Wed, Jul 20, 2011 at 10:27 AM, Gergely Nagy wrote: Not to mention that users who customised their init scripts will suddenly have to figure out how to do the same stuff with systemd - with no automatic upgrade path. No, existing init scripts work with systemd, as stated already in this thread. Otherwise no-one would ever want to try or adopt it. For example, many programs on my system have a file in /etc/default, which file is sourced by the init script. These customisation options will need to be migrated to systemd aswell, which is yet another burden on the Debian package maintainer. Lennart's latest blog post systemd for Administrators, Part IX covers this topic. There even seems to be ways to keep the existing /etc/default files, but personally I would like to see /etc/default go away entirely. http://0pointer.de/blog/projects/on-etc-sysinit.html -- bye, pabs http://wiki.debian.org/PaulWise -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/CAKTje6Ghbo57XzL5uBGWHAubCEEph9bOO2Ow8ik1asyoY3=+x...@mail.gmail.com
Re: [Lennart Poettering] Re: A few observations about systemd
Le mardi 19 juillet 2011 à 21:26 +0200, Wouter Verhelst a écrit : kFreeBSD is currently released as a technology preview. With that, we mean it works, but it isn't necessarily ready yet for prime usage. The fact that there currently aren't many users yet isn't surprising in that light. However, as the port matures, it's not unthinkable that there will be many more users. Frankly, I'd be somewhat surprised if some time after the release of wheezy (but still before wheezy+1), usage of Debian kFreeBSD did not surpass that of the i386 port. You must be joking. So far only a very small subset of packages have been actually checked as working on kfreebsd. People are not going to risk seeing a dysfunctional system just so that they have pf and ZFS, unless they really need it. You disagree with systemd service files being much simpler than sysv init scripts for many daemons? No, I disagree with your statement that life of many maintainers of daemon packages is a complete nightmare currently. Perhaps some things would be a bit easier to do with systemd unit files; but most initscripts are slightly-modified template initscripts from dh-make anyway; it's not as if they require a lot of effort to maintain. It’s not such a lot of effort indeed, but the quality of the result is not impressive either. Our init scripts are overall inconsistent and buggy. -- .''`. Josselin Mouette : :' : `. `' `- -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/1311151866.4372.249.camel@pi0307572
Re: [Lennart Poettering] Re: A few observations about systemd
* Uoti Urpala uoti.urp...@pp1.inet.fi [110719 23:31]: Wouter Verhelst wouter at debian.org writes: Debian is the 'Universal' operating system, and many of our developers (including myself) pride themselves on that. We port to many architectures, we port to multiple kernels. It's one of the defining features of Debian: you can run it /anywhere/ This is an almost religious argument. You take the value of running on multiple kernels as an article of faith, with no evaluation of the benefits (either to people directly using such ports, or possible feedback to the main distribution). It's hard to make rational arguments against such a position, other than to note that the position is irrational and causes practical harm where it interferes with rational decisions. If you do not address the issues, but try to reduce arguments to something almost absurd then of course you will have problems to refute things. Universal is not so much about choice of kernels. It's about not excluding people. Saying This is no problem for 95% of people, why care about the rest if it makes things harder for such a vast majority might sound reasonable if not thinking about it. Of course such decisions will often not be independent, but in that case 14 such decisions would already be enough to rule out over more than half the people. We should care about niches or minorities, if only because every single person in earth is in some niche or part of some minority. Noone if mayority in every single aspect. Please understand that a You are all doing it wrong since ever, everything you know shall no longer have any value, I know how things should be done instead and I do not even care about the big obvious cases where this no longer works will not become sounding better to people by claiming that the obvious victims have no value and should not exist anyway. Bernhard R. Link -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110720090317.ga15...@pcpool00.mathematik.uni-freiburg.de
Re: [Lennart Poettering] Re: A few observations about systemd
On Tue, Jul 19, 2011 at 09:26:33PM +0200, Wouter Verhelst wrote: Frankly, I'd be somewhat surprised if some time after the release of wheezy (but still before wheezy+1), usage of Debian kFreeBSD did not surpass that of the i386 port. I'd be *very* surprised. I can only imagine this happening if the vast majority of Debian users finally moved off to other distributions (i.e., parity by decreasing i386 usage, not be increasing kFreeBSD usage). -- Jon Dowland -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110720091100.GB3796@pris
Re: [Lennart Poettering] Re: A few observations about systemd
On Wed, Jul 20, 2011 at 12:52:36AM +0200, Wouter Verhelst wrote: Where it came from is less important than what it represents today to some of us. I believe I can read in your post that you don't like it; but certainly this is not true for all of us. It completely predates Debian releasing non-Linux kernels and is not mentioned in the social contract. That some people feel it justifies (or even mandates) non-Linux kernels in Debian is a retcon. pf, ZFS; these are valid reasons stated that support kFreeBSD. I interpret 'the Universal OS to mean'… is not. -- Jon Dowland -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110720091529.GC3796@pris
Re: [Lennart Poettering] Re: A few observations about systemd
On Wed, Jul 20, 2011 at 09:37:07AM +0200, Raphael Hertzog wrote: I am among the people who are proud to see that we managed to achieve Debian kfreebsd. Same here, I've always mentioned GNU/kFreeBSD as one of the things I'm most proud of in the Squeeze release. I'd be no less proud of seeing Debian grow other non-Linux ports (e.g. Hurd) in future releases. No matter how unimportant those ports might seem to people using only GNU/Linux, the make Debian today one of the most important porting platforms for our upstreams and users. They can both benefit from Debian seeing how portable their software really is. This is an important role that Debian is playing in the whole ecosystem of Free Software. We should be shaping the future and not be simple followers. I agree with Joey Hess that we should have init systems that make use of all the powers of Linux on Linux and make use of all the powers of FreeBSD on FreeBSD. I agree with this as well. Even if it might seem at stake with the former argument, I believe it is not. We cannot hold back advancements just because one part of our huge archive does not support them; doing so would mean taking a rather extreme (and wrong, imo) side in the trade-off among universality (in the technical sense of runs everywhere) and advanced (when compared to other distros). If we lag behind in features that are good for GNU/Linux users (who are the vast majority of our users) just because users of some ports can't have them, we might force users to choose other distros, renouncing to some of the unique features that Debian has to offer (freedom, quality, open development, etc.). This of course goes both way: we should not hold back non-Linux features on non-Linux kernels because the Linux kernel lack them. Adopting that as a general principle would mean offering, overall, the intersection of features available in all our ports, something which is doomed to reduce with the growth of the number of ports. But what I find surprising in this discussion (with notable exception, luckily) is the feeling that portability is boolean: it is not. It is rather a trade-off among the work that needs to be done / code that needs to be maintained and the distro-wide technical choices that we make. In that respect, the fact that systemd upstream might decide not to integrate upstream our chances is sad, but it's not the end of the world: it won't be the first nor the last upstream not willing to integrate some of our changes. This is why interfaces are much more important than the individual implementations. This is what has been suggested in this thread And speaking about interface, another surprising absence in this thread is the mention of Debian's most important interface, namely the Debian Policy. No matter how much we discuss whether systemd (or upstart, fwiw) is good or not in this thread, the discussion won't make adopting it any easier. init.d scripts are explicitly supported by the Debian Policy and required for packages shipping services. That means that the first mandatory step to have support for a non SysV init system in Debian is to add support for it into policy. That has started after the upstart in Debian BoF at DebConf10 and is being tracked in #591791. I've pointed the systemd maintainer to it a long time ago and he has chimed in there (thanks Tollef!). I'm not following the bug log closely, but it seems to me that they are also discussing there how to generalize the policy change to other init systems, such as systemd. That is very good and has way more chances of changing the status quo in Debian than any pro- or against-systemd thread on -devel. Cheers. -- Stefano Zacchiroli -o- PhD in Computer Science \ PostDoc @ Univ. Paris 7 zack@{upsilon.cc,pps.jussieu.fr,debian.org} -- http://upsilon.cc/zack/ Quando anche i santi ti voltano le spalle, | . |. I've fans everywhere ti resta John Fante -- V. Capossela ...| ..: |.. -- C. Adams signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
On Jul 20, Russ Allbery r...@debian.org wrote: Again, why? ZFS is a pretty big one. It is about as stable as BTRFS on Linux, so I do not see either a compelling argument right now. -- ciao, Marco signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
On Wed, Jul 20, 2011 at 04:31:24PM +0200, Marco d'Itri wrote: On Jul 20, Russ Allbery r...@debian.org wrote: Again, why? ZFS is a pretty big one. It is about as stable as BTRFS on Linux, so I do not see either a compelling argument right now. BTRFS ? stable ? You must be living in the future. Mike -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110720143901.ga9...@glandium.org
Re: [Lennart Poettering] Re: A few observations about systemd
On Jul 20, Mike Hommey m...@glandium.org wrote: Again, why? ZFS is a pretty big one. It is about as stable as BTRFS on Linux, so I do not see either a compelling argument right now. BTRFS ? stable ? You must be living in the future. My point. -- ciao, Marco signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
On Wed, Jul 20, 2011 at 04:31:24PM +0200, Marco d'Itri wrote: On Jul 20, Russ Allbery r...@debian.org wrote: Again, why? ZFS is a pretty big one. It is about as stable as BTRFS on Linux, so I do not see either a compelling argument right now. Do you have any data to back up that statement? There have been some issues in the past with ZFS on 32bit machines and/or low-memory situations, but if you use it in the environment that it's meant for (large deployments), it's pretty stable. -- The volume of a pizza of thickness a and radius z can be described by the following formula: pi zz a -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110720144830.ge5...@grep.be
Re: [Lennart Poettering] Re: A few observations about systemd
m...@linux.it (Marco d'Itri) writes: On Jul 20, Russ Allbery r...@debian.org wrote: ZFS is a pretty big one. It is about as stable as BTRFS on Linux, so I do not see either a compelling argument right now. I know from actual, real-world testing and usage of specifically Debian kFreeBSD that ZFS is stable enough for production use. I don't personally know anything about BTRFS, so I don't know if it's a comprehensive replacement for the facilities of ZFS, or anything about its stability. -- Russ Allbery (r...@debian.org) http://www.eyrie.org/~eagle/ -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87r55kzz6d@windlord.stanford.edu
Re: [Lennart Poettering] Re: A few observations about systemd
On Mon, Jul 18, 2011 at 04:03:41PM -0700, Steve Langasek wrote: On Mon, Jul 18, 2011 at 11:14:39PM +0200, Mike Hommey wrote: I'm sure that systemd does much better than a traditional sysvinit boot with /bin/bash and no dependency-based booting. But then, so does Debian's current boot system, and so does upstart; and neither of the latter two involve grandiose claims of a shell-free boot. Trying to take the shell completely out of the boot means a definite tradeoff here between boot speed and configurability/maintainability, and in the absence of hard numbers, I suspect this is a false optimization and not a trade-off that we actually want to make in a general distribution. Which then calls into question the use of such claims as a justification for a switch to systemd at all... I'd expect some important differences between shell script based init and systemd-type init by the simple fact that there are (or at least should be, I don't know how systemd actually works) less files to read. Less files to read == less disk seeks. Disks seeks hurt startup performance. Yes, I've read your blog entries on the subject. :-) That's true, but I think the reduction in the number of files being accessed, for systemd vs. sysvinit or upstart, is rather small; aside from some things in /etc/rcS.d, most init scripts would have approximately a 1:1 correlation with upstart jobs or systemd config files, and if you've read the shell off disk once it's in cache and there's not likely to be any more seeking. So I do expect that most of the shell penalty will be CPU rather than disk in the context of boot. But you don't only load the /etc/rcS.d scripts, you also need to load all the executables the scripts use, and their dependent libraries, and the files they need to read. But yeah, that would still need to be accounted and compared. Mike -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110719062511.ga3...@glandium.org
Re: [Lennart Poettering] Re: A few observations about systemd
On Mon, Jul 18, 2011 at 11:05:56PM +, Uoti Urpala wrote: I think the important question is whether portability to other kernels is or should be a project's goal, and how much else you're willing to lose for the sake of that goal. Debian/kFreeBSD is here to stay, it's not going away. With that as a given, systemd is suddenly a lot less interesting. Whatever its features, if we have to jump through a large heap of hoops to get it to work at all, or to make life for maintainers of daemon packages not a complete nightmare, it's not likely to become the default in Debian any time soon. -- The volume of a pizza of thickness a and radius z can be described by the following formula: pi zz a -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110719092835.gq2...@grep.be
Re: [Lennart Poettering] Re: A few observations about systemd
On Mon, Jul 18, 2011 at 11:48:35PM +, Uoti Urpala wrote: By the way, I think in exchange for faster boot is focusing too narrowly on boot speed. It's not like boot speed would be the only reason to avoid shell. You do realize that you're talking to a mailinglist populated mostly by people who spend most of their free time writing shell scripts, right? -- The volume of a pizza of thickness a and radius z can be described by the following formula: pi zz a -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110719093135.gr2...@grep.be
Re: [Lennart Poettering] Re: A few observations about systemd
It's not like boot speed would be the only reason to avoid shell. I don't think that avoiding shell implies that all the distribution- specific initialisation code must be hard-wired in pid 1. I'd be more sympathetic to the idea of recoding everything in C if the initiali- sation code lived in separate binaries. -- Juliusz -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87mxgah4l1@trurl.pps.jussieu.fr
Re: [Lennart Poettering] Re: A few observations about systemd
]] Juliusz Chroboczek | It's not like boot speed would be the only reason to avoid shell. | | I don't think that avoiding shell implies that all the distribution- | specific initialisation code must be hard-wired in pid 1. I'd be more | sympathetic to the idea of recoding everything in C if the initiali- | sation code lived in separate binaries. Like: : tfheen@qurzaw /lib/systemd ls system/ systemd-fsck* systemd-quotacheck* systemd-shutdown* systemd-vconsole-setup* systemd-ac-power* systemd-hostnamed* systemd-random-seed* systemd-shutdownd* system-generators/ systemd-binfmt* systemd-initctl* systemd-readahead-collect* systemd-sysctl* system-shutdown/ systemd-cgroups-agent* systemd-kmsg-syslogd* systemd-readahead-replay* systemd-timestamp* systemd-cryptsetup* systemd-logger*systemd-remount-api-vfs* systemd-update-utmp* systemd-detect-virt*systemd-modules-load* systemd-reply-password* systemd-user-sessions* ? Really, it's not that hard coded. Regards, -- Tollef Fog Heen UNIX is user friendly, it's just picky about who its friends are -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87vcuyh3cz@qurzaw.varnish-software.com
Re: [Lennart Poettering] Re: A few observations about systemd
| I'd be more sympathetic to the idea of recoding everything in C if | the initialiation code lived in separate binaries. system/ systemd-fsck* systemd-quotacheck* systemd-shutdown* systemd-vconsole-setup* [...] Interesting. Looking at the code, I hadn't noticed these get compiled into separate utilities. Really, it's not that hard coded. Indeed -- they are simply ExecStart:ed by normal systemd units. -- Juliusz -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87fwm2h28g@trurl.pps.jussieu.fr
Re: [Lennart Poettering] Re: A few observations about systemd
Wouter Verhelst wouter at debian.org writes: On Mon, Jul 18, 2011 at 11:05:56PM +, Uoti Urpala wrote: I think the important question is whether portability to other kernels is or should be a project's goal, and how much else you're willing to lose for the sake of that goal. Debian/kFreeBSD is here to stay, it's not going away. With that as a given, systemd is suddenly a lot less interesting. Once you stop taking things as a given there are a lot more opportunities for improvement. If you want to do whatever work is necessary to keep kFreeBSD working that's fine of course. But the attitude that it's OK for kFreeBSD to set limits on Linux development (or that developers working on Linux must handle the BSD porting/compatibility to be permitted to adopt a new technology) smells of trying to hold the project hostage, and I doubt it can have positive effects for the project overall. IMO letting kFreeBSD block a technology like systemd (or even letting it have a significant impact on the discussion about whether it's desirable to introduce the technology for the main Linux case) would only be justifiable if there were very solid arguments why kFreeBSD is a big net win for the project, or after a vote showing significant support for the port. Whatever its features, if we have to jump through a large heap of hoops to get it to work at all, or to make life for maintainers of daemon packages not a complete nightmare, it's not likely to become the default in Debian any time soon. I think the life of many maintainers of daemon packages is a complete nightmare now with sysvinit, compared to what it would be with systemd. -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/loom.20110719t144003...@post.gmane.org
Re: [Lennart Poettering] Re: A few observations about systemd
On Tue, Jul 19, 2011 at 01:12:33PM +, Uoti Urpala wrote: Wouter Verhelst wouter at debian.org writes: On Mon, Jul 18, 2011 at 11:05:56PM +, Uoti Urpala wrote: I think the important question is whether portability to other kernels is or should be a project's goal, and how much else you're willing to lose for the sake of that goal. Debian/kFreeBSD is here to stay, it's not going away. With that as a given, systemd is suddenly a lot less interesting. Once you stop taking things as a given there are a lot more opportunities for improvement. kFreeBSD is hardly the only reason why systemd is a bad idea for Debian. If you want to do whatever work is necessary to keep kFreeBSD working that's fine of course. But the attitude that it's OK for kFreeBSD to set limits on Linux development (or that developers working on Linux must handle the BSD porting/compatibility to be permitted to adopt a new technology) smells of trying to hold the project hostage, and I doubt it can have positive effects for the project overall. There's nothing wrong with requiring portability. IMO letting kFreeBSD block a technology like systemd (or even letting it have a significant impact on the discussion about whether it's desirable to introduce the technology for the main Linux case) would only be justifiable if there were very solid arguments why kFreeBSD is a big net win for the project, or after a vote showing significant support for the port. IMAO, a statement of (paraphrased) 'portability is for weenies' isn't doing any of us a favor. Systemd might've been nice if it was portable to other kernels; the fact that it isn't, makes it less than useful for Debian. This would have been okay if upstream would be ready to accept patches. But apparently he's not even willing to consider the possibility of doing so. Yuck. Whatever its features, if we have to jump through a large heap of hoops to get it to work at all, or to make life for maintainers of daemon packages not a complete nightmare, it's not likely to become the default in Debian any time soon. I think the life of many maintainers of daemon packages is a complete nightmare now with sysvinit, compared to what it would be with systemd. It's of course your prerogative to have that opinion, but (as a maintainer of a source package that ships two initscripts) I disagree with it. Especially since I doubt that supporting NBD exports with systemd is going to be possible, at all, given what I know about it. At any rate, if we need to support more than one init system just so that Debian continues to work on more than just Linux, then Something Is Very Wrong(tm). -- The volume of a pizza of thickness a and radius z can be described by the following formula: pi zz a -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110719142122.gx19...@grep.be
Re: [Lennart Poettering] Re: A few observations about systemd
Uoti Urpala uoti.urp...@pp1.inet.fi writes: Whatever its features, if we have to jump through a large heap of hoops to get it to work at all, or to make life for maintainers of daemon packages not a complete nightmare, it's not likely to become the default in Debian any time soon. I think the life of many maintainers of daemon packages is a complete nightmare now with sysvinit, compared to what it would be with systemd. FYI, there are upstreams who provide initscripts in their source package. systemd is yet another burden on them that they have to maintain, and makes their lives miserable. Looking at the syslog-ng sources, there's about a dozen initscripts, including upstart and systemd that we have in git. Now THAT is a pain, a pain that systemd will only make worse, no matter how good it may be. If there was one common init system, that would be awesome, but there isn't, and systemd will never be that system, for reasons that should be obvious to anyone reading this thread, who's not ignoring everything but Linux. In other words, even if systemd can be made portable enough for Debian's needs, or Debian can find a way to work around systemds unportability, upstreams who need to support other systems will still have yet another extra burden to carry. -- |8] -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87hb6ie3rj@balabit.hu
Re: [Lennart Poettering] Re: A few observations about systemd
]] Wouter Verhelst | It's of course your prerogative to have that opinion, but (as a | maintainer of a source package that ships two initscripts) I disagree | with it. Especially since I doubt that supporting NBD exports with | systemd is going to be possible, at all, given what I know about it. I wish people would stop saying stuff like this without actually testing first. nbd-{client,server} works fine with systemd already, and from looking at the init scripts there's absolutely nothing that stops it from being started by a .service file either. Cheers, -- Tollef Fog Heen UNIX is user friendly, it's just picky about who its friends are -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87mxgagu1z@qurzaw.varnish-software.com
Re: [Lennart Poettering] Re: A few observations about systemd
]] Gergely Nagy | FYI, there are upstreams who provide initscripts in their source | package. systemd is yet another burden on them that they have to | maintain, and makes their lives miserable. You make it sound like systemd requires you to make an extra effort to make stuff work. It doesn't. SystemV initscripts work just fine. If you want to do stuff like socket activation and so on you need to make an extra effort, but that is in no way required to use systemd. While the support for sysvinit scripts can be compiled out of systemd I have absolutely no plans to do so for the overseeable future, if ever. Cheers, -- Tollef Fog Heen UNIX is user friendly, it's just picky about who its friends are -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87ipqygtoc@qurzaw.varnish-software.com
Re: [Lennart Poettering] Re: A few observations about systemd
Wouter Verhelst wouter at debian.org writes: Debian/kFreeBSD is here to stay, it's not going away. With that as a given, systemd is suddenly a lot less interesting. Once you stop taking things as a given there are a lot more opportunities for improvement. kFreeBSD is hardly the only reason why systemd is a bad idea for Debian. It's the only argument I've seen you mention. And I don't remember seeing convincing arguments against it from anyone else in the thread either. If you want to do whatever work is necessary to keep kFreeBSD working that's fine of course. But the attitude that it's OK for kFreeBSD to set limits on Linux development (or that developers working on Linux must handle the BSD porting/compatibility to be permitted to adopt a new technology) smells of trying to hold the project hostage, and I doubt it can have positive effects for the project overall. There's nothing wrong with requiring portability. Of course there is when it interferes with other goals. And your claim would at least require a lot of further qualifications to avoid being totally absurd - if you say any portability requirement whatsoever is fine, how much software would remain in Debian after requiring portability to Windows? OS/2? IMO letting kFreeBSD block a technology like systemd (or even letting it have a significant impact on the discussion about whether it's desirable to introduce the technology for the main Linux case) would only be justifiable if there were very solid arguments why kFreeBSD is a big net win for the project, or after a vote showing significant support for the port. IMAO, a statement of (paraphrased) 'portability is for weenies' isn't Keeping portability in mind is a good thing especially if you're doing something that is easily implementable with common interfaces. However in some cases additional portability has very real costs, and it's by no means given that the balance should go in the favor of portability. doing any of us a favor. Systemd might've been nice if it was portable to other kernels; the fact that it isn't, makes it less than useful for Debian. That systemd isn't portable to other kernels affects kFreeBSD and Hurd. Given that the contribution of those to the overall usefulness of Debian is negligible, this has little effect on how useful systemd is for Debian. Whatever its features, if we have to jump through a large heap of hoops to get it to work at all, or to make life for maintainers of daemon packages not a complete nightmare, it's not likely to become the default in Debian any time soon. I think the life of many maintainers of daemon packages is a complete nightmare now with sysvinit, compared to what it would be with systemd. It's of course your prerogative to have that opinion, but (as a maintainer of a source package that ships two initscripts) I disagree with it. You disagree with systemd service files being much simpler than sysv init scripts for many daemons? It may be your prerogative to hold that opinion, but I think it contradicts reality. Especially since I doubt that supporting NBD exports with systemd is going to be possible, at all, given what I know about it. You mean supporting that functionality would be completely impossible on a system running systemd? Really? On what do you base that belief? At any rate, if we need to support more than one init system just so that Debian continues to work on more than just Linux, then Something Is Very Wrong(tm). I don't consider it a big problem if someone uses his prerogative to Feel That Something Is Very Wrong, as long as things work well in practice. -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/loom.20110719t173405-...@post.gmane.org
Re: [Lennart Poettering] Re: A few observations about systemd
Le mardi 19 juillet 2011 à 16:36 +0200, Gergely Nagy a écrit : FYI, there are upstreams who provide initscripts in their source package. And how many of them comply with the Debian policy without needing to be completely rewritten? Let’s talk about real cases, please. -- .''`. Josselin Mouette : :' : `. `' `- -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/1311091336.4372.174.camel@pi0307572
Re: [Lennart Poettering] Re: A few observations about systemd
Gergely Nagy algernon at balabit.hu writes: Uoti Urpala uoti.urpala at pp1.inet.fi writes: Whatever its features, if we have to jump through a large heap of hoops to get it to work at all, or to make life for maintainers of daemon packages not a complete nightmare, it's not likely to become the default in Debian any time soon. I think the life of many maintainers of daemon packages is a complete nightmare now with sysvinit, compared to what it would be with systemd. FYI, there are upstreams who provide initscripts in their source package. systemd is yet another burden on them that they have to maintain, and makes their lives miserable. It's one of the goals of systemd to allow more upstreams to provide reasonable init scripts. Adding systemd may create more work for upstreams that insist on providing a suitable script for every possible distro, but on the other hand adoption of systemd makes it easier for more upstreams to provide a reasonable service configuration that works on most systems (systemd configuration is a lot more portable than a sysv init script). -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/loom.20110719t180255-...@post.gmane.org
Re: [Lennart Poettering] Re: A few observations about systemd
Tollef Fog Heen tfh...@err.no writes: ]] Gergely Nagy | FYI, there are upstreams who provide initscripts in their source | package. systemd is yet another burden on them that they have to | maintain, and makes their lives miserable. You make it sound like systemd requires you to make an extra effort to make stuff work. It doesn't. SystemV initscripts work just fine. If you want to do stuff like socket activation and so on you need to make an extra effort, but that is in no way required to use systemd. While the support for sysvinit scripts can be compiled out of systemd I have absolutely no plans to do so for the overseeable future, if ever. If not using the goodies provided by systemd, there's absolutely no point in writing a service file to use with it, imo. And as you wrote, socket activation does need extra effort. Not just in the service file, but in the code aswell. Yes, upstream can choose to ignore systemd, just like systemd upstream choses to ignore anything non-Linux. But that's not going to make users happy. It will make distributions that default to systemd, and have the daemon in question (syslog-ng) in their default install even less happy. At that point, there really is no choice but to go ahead and do the extra effort. It wasn't trivial, and now we need to take extra care not to break our systemd support. Hello, extra maintainance burden! (Do note, that I like some of the stuff systemd can bring to the table, and I belive that going the extra mile in syslog-ng's case was worth the trouble, but it's still extra code we'll have to carry. Much like the AIX and Solaris workarounds.) -- |8] -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/877h7egq35@luthien.mhp
Re: [Lennart Poettering] Re: A few observations about systemd
Uoti Urpala uoti.urp...@pp1.inet.fi writes: Gergely Nagy algernon at balabit.hu writes: Uoti Urpala uoti.urpala at pp1.inet.fi writes: Whatever its features, if we have to jump through a large heap of hoops to get it to work at all, or to make life for maintainers of daemon packages not a complete nightmare, it's not likely to become the default in Debian any time soon. I think the life of many maintainers of daemon packages is a complete nightmare now with sysvinit, compared to what it would be with systemd. FYI, there are upstreams who provide initscripts in their source package. systemd is yet another burden on them that they have to maintain, and makes their lives miserable. It's one of the goals of systemd to allow more upstreams to provide reasonable init scripts. Adding systemd may create more work for upstreams that insist on providing a suitable script for every possible distro, but on the other hand adoption of systemd makes it easier for more upstreams to provide a reasonable service configuration that works on most systems (systemd configuration is a lot more portable than a sysv init script). systemd is Yet Another init system to support, however you put it. And if one wants to move out of the 1970's (as someone in this thread said it earlier), and make use of systemd features where available, then it can also result in non-trivial amount of extra code to carry in the code base, on top of the service file. Thus, upstream has to jump through a large heap of hoops to support systemd properly (and if not going for proper systemd support, making use of its new features, I see no point in writing a service file to begin with). Been there, done that, wasn't as easy as writing a service file. -- |8] -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87zkkafb2t@luthien.mhp
Re: [Lennart Poettering] Re: A few observations about systemd
On Tue, Jul 19, 2011 at 07:31:59AM +0200, Tollef Fog Heen wrote: | Also, I've installed systemd on my laptop and it logs almost nothing | to the console (verbose on the kernel command line does not help). try doing systemd.log_level=debug as documented in the man page? It's hard to access the manpage on the only system on which systemd is installed when the system won't boot. If everything's working normally, I don't care if it's mostly silent. -- brian m. carlson / brian with sandals: Houston, Texas, US +1 832 623 2791 | http://www.crustytoothpaste.net/~bmc | My opinion only OpenPGP: RSA v4 4096b: 88AC E9B2 9196 305B A994 7552 F1BA 225C 0223 B187 signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
[Uoti Urpala] IMO letting kFreeBSD block a technology like systemd (or even letting it have a significant impact on the discussion about whether it's desirable to introduce the technology for the main Linux case) would only be justifiable if there were very solid arguments why kFreeBSD is a big net win for the project, or after a vote showing significant support for the port. IMO letting systemd block a technology like kFreeBSD (or even letting it have a significant impact on the discussion about whether it's desirable to introduce the port for Debian releases) would only be justifiable if there were very solid arguments why systemd is a big net win for the project, or after a vote showing significant support for the package. Just sayin', -- Peter Samuelson | org-tld!p12n!peter | http://p12n.org/ -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110719184104.gg4...@p12n.org
Re: [Lennart Poettering] Re: A few observations about systemd
On Tue, Jul 19, 2011 at 3:41 PM, Peter Samuelson pe...@p12n.org wrote: [Uoti Urpala] IMO letting kFreeBSD block a technology like systemd (or even letting it have a significant impact on the discussion about whether it's desirable to introduce the technology for the main Linux case) would only be justifiable if there were very solid arguments why kFreeBSD is a big net win for the project, or after a vote showing significant support for the port. IMO letting systemd block a technology like kFreeBSD (or even letting it have a significant impact on the discussion about whether it's desirable to introduce the port for Debian releases) would only be justifiable if there were very solid arguments why systemd is a big net win for the project, or after a vote showing significant support for the package. What I don't understand is, why do so many people think they're mutually exclusive? I guess this thread is past the point where nothing useful can come out of it anymore. -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/canvyna-myrysmfrzf1hoz9ue_phcbbtfursp_ccbfecbiil...@mail.gmail.com
Re: [Lennart Poettering] Re: A few observations about systemd
On Tue, Jul 19, 2011 at 03:59:13PM +, Uoti Urpala wrote: Wouter Verhelst wouter at debian.org writes: Debian/kFreeBSD is here to stay, it's not going away. With that as a given, systemd is suddenly a lot less interesting. Once you stop taking things as a given there are a lot more opportunities for improvement. kFreeBSD is hardly the only reason why systemd is a bad idea for Debian. It's the only argument I've seen you mention. And I don't remember seeing convincing arguments against it from anyone else in the thread either. Pfff. You're the one who wants to change the status quo, not me. If you want to do whatever work is necessary to keep kFreeBSD working that's fine of course. But the attitude that it's OK for kFreeBSD to set limits on Linux development (or that developers working on Linux must handle the BSD porting/compatibility to be permitted to adopt a new technology) smells of trying to hold the project hostage, and I doubt it can have positive effects for the project overall. There's nothing wrong with requiring portability. Of course there is when it interferes with other goals. And your claim would at least require a lot of further qualifications to avoid being totally absurd - It's always possible to read absurdity in a totally reasonable statement; that doesn't make the original statement absurd. if you say any portability requirement whatsoever is fine, I said nothing of the sort. Currently, Debian requires that software in the Essential set of packages be portable to any of our kernels. It does not even require POSIX, let alone any portability requirement whatsoever. [...] IMO letting kFreeBSD block a technology like systemd (or even letting it have a significant impact on the discussion about whether it's desirable to introduce the technology for the main Linux case) would only be justifiable if there were very solid arguments why kFreeBSD is a big net win for the project, or after a vote showing significant support for the port. IMAO, a statement of (paraphrased) 'portability is for weenies' isn't Keeping portability in mind is a good thing especially if you're doing something that is easily implementable with common interfaces. Init has been implemented since the early 70s. I'd say that qualifies it as easily implementable with common interfaces. Of course it's true that there are issues with the current init implementation, and that a replacement would be nice. And yes, of course it's true that one can cut corners by focussing on just one kernel, and not caring about the rest; that way, you can get a working implementation quickly with much less effort than you would if you were to keep portability a requirement from the very beginning. However, that doesn't mean it's the only possible way, or indeed the most desirable one. [...] doing any of us a favor. Systemd might've been nice if it was portable to other kernels; the fact that it isn't, makes it less than useful for Debian. That systemd isn't portable to other kernels affects kFreeBSD and Hurd. Given that the contribution of those to the overall usefulness of Debian is negligible, this has little effect on how useful systemd is for Debian. kFreeBSD is currently released as a technology preview. With that, we mean it works, but it isn't necessarily ready yet for prime usage. The fact that there currently aren't many users yet isn't surprising in that light. However, as the port matures, it's not unthinkable that there will be many more users. Frankly, I'd be somewhat surprised if some time after the release of wheezy (but still before wheezy+1), usage of Debian kFreeBSD did not surpass that of the i386 port. [...] Whatever its features, if we have to jump through a large heap of hoops to get it to work at all, or to make life for maintainers of daemon packages not a complete nightmare, it's not likely to become the default in Debian any time soon. I think the life of many maintainers of daemon packages is a complete nightmare now with sysvinit, compared to what it would be with systemd. It's of course your prerogative to have that opinion, but (as a maintainer of a source package that ships two initscripts) I disagree with it. You disagree with systemd service files being much simpler than sysv init scripts for many daemons? No, I disagree with your statement that life of many maintainers of daemon packages is a complete nightmare currently. Perhaps some things would be a bit easier to do with systemd unit files; but most initscripts are slightly-modified template initscripts from dh-make anyway; it's not as if they require a lot of effort to maintain. [...rest of this trolling snipped...] -- The volume of a pizza of thickness a and radius z can be described by the following formula: pi zz a signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
On Jul 19, Wouter Verhelst wou...@debian.org wrote: kFreeBSD is currently released as a technology preview. With that, we mean it works, but it isn't necessarily ready yet for prime usage. The fact that there currently aren't many users yet isn't surprising in that light. However, as the port matures, it's not unthinkable that there will be many more users. Again, why? So far we have one person who likes pf and one who suspects that maybe FreeBSD could behave better when severely overloaded. Frankly, I'd be somewhat surprised if some time after the release of wheezy (but still before wheezy+1), usage of Debian kFreeBSD did not surpass that of the i386 port. And do you have any data to justify this prediction or did you see it in the crystal ball? -- ciao, Marco signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
On Tue, Jul 19, 2011 at 09:38:04PM +0200, Marco d'Itri wrote: On Jul 19, Wouter Verhelst wou...@debian.org wrote: kFreeBSD is currently released as a technology preview. With that, we mean it works, but it isn't necessarily ready yet for prime usage. The fact that there currently aren't many users yet isn't surprising in that light. However, as the port matures, it's not unthinkable that there will be many more users. Again, why? So far we have one person who likes pf and one who suspects that maybe FreeBSD could behave better when severely overloaded. Oh, I get it, we should throw kFreeBSD out because Marco d'Itri thinks it's a bad idea? In less than a day, you've had two people come up with, for them, good arguments. Another oft-quoted argument for kFreeBSD is ZFS (which does indeed have some very nice features). Granted, this isn't a scientifically valid poll, but come on, it's a troll thread on a development mailinglist, what do you expect? Frankly, I'd be somewhat surprised if some time after the release of wheezy (but still before wheezy+1), usage of Debian kFreeBSD did not surpass that of the i386 port. And do you have any data to justify this prediction or did you see it in the crystal ball? Call it gut feeling. -- The volume of a pizza of thickness a and radius z can be described by the following formula: pi zz a -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110719194710.gc6...@grep.be
Re: [Lennart Poettering] Re: A few observations about systemd
Peter Samuelson peter at p12n.org writes: [Uoti Urpala] IMO letting kFreeBSD block a technology like systemd (or even letting it have a significant impact on the discussion about whether it's desirable to introduce the technology for the main Linux case) would only be justifiable if there were very solid arguments why kFreeBSD is a big net win for the project, or after a vote showing significant support for the port. IMO letting systemd block a technology like kFreeBSD (or even letting it have a significant impact on the discussion about whether it's desirable to introduce the port for Debian releases) would only be justifiable if there were very solid arguments why systemd is a big net win for the project, or after a vote showing significant support for the package. Are the differences really so hard to understand? Well, since you seem to honestly believe that would work as an argument I'll try to explain. Even if systemd did not exist at all, kFreeBSD would not have become a core Debian technology. There was a discussion about whether future Debian would be based on kFreeBSD, and kFreeBSD failed that on its own merits, not due to any consideration of systemd (or actually there wasn't much of a discussion, but that was only because kFreeBSD's failure was so obvious). Future Debian will not be based kFreeBSD, nor will it even be an important alternative. Linux is the core of Debian, what almost all users use and where most of the development that keeps Debian running happens. It's clear that kFreeBSD will have very little if any positive contribution to that core. systemd deserves to be evaluated on its overall benefit, without giving undue weight to irrelevancies like kFreeBSD. And if it's determined to be beneficial on Linux, then harm to the few potential users of Debian/kFreeBSD is irrelevant compared to the majority using Linux. Surely you could find MUCH bigger groups of users who could benefit with a smaller amount of effort. kFreeBSD is also mainly a developer toy with little general user demand. Using the existence of such a toy as a reason to block other developers from working on systemd integration would be especially questionable: you'd keep other developers from working on something that benefits the majority of users, for the sake of a toy that exists primarily for the developer's own amusement. -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/loom.20110719t211619-...@post.gmane.org
Re: [Lennart Poettering] Re: A few observations about systemd
* Russ Allbery (r...@debian.org) [110719 01:36]: Uoti Urpala uoti.urp...@pp1.inet.fi writes: I know I would personally be a lot happier with a Debian that supports systemd functionality than with a Debian that can run on a BSD kernel. Well, while we're putting stakes in the ground, I suppose I'll hammer mine in there as well. I completely disagree to the point that I would take that to a GR. The decision is already taken that Debian can run on BSD kernels. So if someone wants to revert that decision, it'd need an GR. Not the other way. Andi -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110719195235.gz15...@mails.so.argh.org
Re: [Lennart Poettering] Re: A few observations about systemd
On Tue, 2011-07-19 at 19:48 +, Uoti Urpala wrote: There was a discussion about whether future Debian would be based on kFreeBSD, and kFreeBSD failed that on its own merits, not due to any consideration of systemd (or actually there wasn't much of a discussion, but that was only because kFreeBSD's failure was so obvious). Future Debian will not be based kFreeBSD, nor will it even be an important alternative. Please share more details of this discussion to which you refer. Whilst doing so, it might be worth bearing in mind that I'm (at least) the second member of the release team to reply in this thread and that there are no current plans for kfreebsd-{amd64,i386} to cease being release architectures for Debian. Regards, Adam -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/1311105733.8028.10.ca...@hathi.jungle.funky-badger.org
Re: [Lennart Poettering] Re: A few observations about systemd
On Jul 19, Wouter Verhelst wou...@debian.org wrote: Oh, I get it, we should throw kFreeBSD out because Marco d'Itri thinks it's a bad idea? No, but if you believe it to be useful the least you could do is to explain why. -- ciao, Marco signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
Wouter Verhelst wouter at debian.org writes: kFreeBSD is hardly the only reason why systemd is a bad idea for Debian. It's the only argument I've seen you mention. And I don't remember seeing convincing arguments against it from anyone else in the thread either. Pfff. You're the one who wants to change the status quo, not me. So Pfff is a valid answer to all attempts to change the status quo? There have been arguments for systemd (in this thread, or if you really can't find any, go look at upstream website for example). Those have not been refuted. There have not been good arguments against it. Now your level of argumentation is to say but it's a bad idea, and Pfff? Do you really not have anything less inane to contribute? If you want to do whatever work is necessary to keep kFreeBSD working that's fine of course. But the attitude that it's OK for kFreeBSD to set limits on Linux development (or that developers working on Linux must handle the BSD porting/compatibility to be permitted to adopt a new technology) smells of trying to hold the project hostage, and I doubt it can have positive effects for the project overall. There's nothing wrong with requiring portability. Of course there is when it interferes with other goals. And your claim would at least require a lot of further qualifications to avoid being totally absurd - It's always possible to read absurdity in a totally reasonable statement; that doesn't make the original statement absurd. Your original statement perhaps looks totally reasonable at first sight. But it is not reasonable and its vagueness helps hide that. if you say any portability requirement whatsoever is fine, I said nothing of the sort. Currently, Debian requires that software in the Essential set of packages be portable to any of our kernels. It does not even require If you had phrased your original statement as There's nothing wrong with requiring portability to exactly the set of kernels that I want to consider as official Debian kernels, and which set of kernels is not subject to re-evaluation. then it wouldn't look quite so totally reasonable would it? Of course it's true that there are issues with the current init implementation, and that a replacement would be nice. And yes, of course it's true that one can cut corners by focussing on just one kernel, and not caring about the rest; that way, you can get a working implementation quickly with much less effort than you would if you were to keep portability a requirement from the very beginning. However, that doesn't mean it's the only possible way, or indeed the most desirable one. And at which point would it become desirable? I think systemd upstream's statements about saving significant amounts of work by not considering a larger set of OSes are quite plausible. It's easy to say that you want portability too. And a free pony with every copy. But that work would have been lost from something else, and the result would likely be significantly worse overall. Even if you disagree about systemd upstream's views on portability that does not change the quality of the software and how it performs on Linux. IMO attitudes like if upstream holds such heretical views then their software is not fit for use on any platform are not justified. -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/loom.20110719t215901-...@post.gmane.org
Re: [Lennart Poettering] Re: A few observations about systemd
On Tue, Jul 19, 2011 at 08:25:19PM +, Uoti Urpala wrote: Even if you disagree about systemd upstream's views on portability that does not change the quality of the software and how it performs on Linux. IMO attitudes like if upstream holds such heretical views then their software is not fit for use on any platform are not justified. Debian is the 'Universal' operating system, and many of our developers (including myself) pride themselves on that. We port to many architectures, we port to multiple kernels. It's one of the defining features of Debian: you can run it /anywhere/ In that light, an init replacement that is non-portable is not quite interesting. I didn't say systemd is not fit for use on any platform. I just said it's not fit for Debian. -- The volume of a pizza of thickness a and radius z can be described by the following formula: pi zz a signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
* Marco d'Itri m...@linux.it [2011-07-19 21:39]: So far we have one person who likes pf and one who suspects that maybe FreeBSD could behave better when severely overloaded. So if trolling is on add another one: it doesn't have udev Kthxgoodby -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110719203029.gn19...@anguilla.debian.or.at
Re: [Lennart Poettering] Re: A few observations about systemd
]] brian m. carlson Hi, | On Tue, Jul 19, 2011 at 07:31:59AM +0200, Tollef Fog Heen wrote: | | Also, I've installed systemd on my laptop and it logs almost nothing | | to the console (verbose on the kernel command line does not help). | | try doing systemd.log_level=debug as documented in the man page? | | It's hard to access the manpage on the only system on which systemd is | installed when the system won't boot. If everything's working | normally, I don't care if it's mostly silent. Then install it somewhere else? It's not like it has much in the way of extra deps, nor does it take over your boot unless you install systemd-sysv. Or just boot with init=/bin/sh and the check the man page. Really, I'm sure you can work it out. I'm sure you can even find the man page on the internet if you try. Cheers, -- Tollef Fog Heen UNIX is user friendly, it's just picky about who its friends are -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87ei1mgflb@qurzaw.varnish-software.com
Re: [Lennart Poettering] Re: A few observations about systemd
Adam D. Barratt adam at adam-barratt.org.uk writes: On Tue, 2011-07-19 at 19:48 +, Uoti Urpala wrote: There was a discussion about whether future Debian would be based on kFreeBSD, and kFreeBSD failed that on its own merits, not due to any consideration of systemd (or actually there wasn't much of a discussion, but that was only because kFreeBSD's failure was so obvious). Future Debian will not be based kFreeBSD, nor will it even be an important alternative. Please share more details of this discussion to which you refer. Whilst doing so, it might be worth bearing in mind that I'm (at least) the second member of the release team to reply in this thread and that there are no current plans for kfreebsd-{amd64,i386} to cease being release architectures for Debian. From the part you already quoted: or actually there wasn't much of a discussion, but that was only because kFreeBSD's failure was so obvious. You seem to interpret Debian would be based on kFreeBSD as we have some official support for kFreeBSD, but that's not what I was talking about. I know there have been official Debian/kFreeBSD releases, but those make very little difference - someone published a set of files which were then ignored by about everyone. I'm talking about what kernel people use when they use Debian, and what platform development that creates the software in the distribution happens on. There's no plan to migrate from Linux to BSD kernel (and very little chance of such a migration happening to any significant degree without an explicit decision or consensus as a result of just having a kFreeBSD release available). By contrast releasing Linux with systemd would mean actual widespread use and would have a practical effect for a large number of people. So the comparison of systemd blocks kFreeBSD vs kFreeBSD blocks systemd that I was originally replying to was comparing very different things. -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/loom.20110719t222744-...@post.gmane.org
Re: [Lennart Poettering] Re: A few observations about systemd
Andreas Barth wrote: The decision is already taken that Debian can run on BSD kernels. So if someone wants to revert that decision, it'd need an GR. Not the other way. That decision was made without a GR, and can manifestly be reversed without a GR. Otherwise the release team's architecture qualification stuff, and the ftp team's architecture removal stuff wouldn't make much sense. -- see shy jo signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
On Jul 19, Martin Wuertele m...@debian.org wrote: So far we have one person who likes pf and one who suspects that maybe FreeBSD could behave better when severely overloaded. So if trolling is on add another one: it doesn't have udev Hint: just because you cannot answer a question, it is not trolling. -- ciao, Marco signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
On Tue, 2011-07-19 at 20:51 +, Uoti Urpala wrote: Adam D. Barratt adam at adam-barratt.org.uk writes: On Tue, 2011-07-19 at 19:48 +, Uoti Urpala wrote: There was a discussion about whether future Debian would be based on kFreeBSD, and kFreeBSD failed that on its own merits, not due to any consideration of systemd (or actually there wasn't much of a discussion, but that was only because kFreeBSD's failure was so obvious). Future Debian will not be based kFreeBSD, nor will it even be an important alternative. Please share more details of this discussion to which you refer. [...] From the part you already quoted: or actually there wasn't much of a discussion, but that was only because kFreeBSD's failure was so obvious. I wasn't going to reply to this (and may wish I hadn't), but... Proof by assertion isn't an argument. If you think kfreebsd sucks then you're entitled to that opinion, but please don't seek to frame it as some sort of consensus direction on the part of the project because it's obvious. You seem to interpret Debian would be based on kFreeBSD as we have some official support for kFreeBSD, but that's not what I was talking about. I know there have been official Debian/kFreeBSD releases, but those make very little difference - someone published a set of files which were then ignored by about everyone. fwiw, they're more official than the original amd64 release of Debian was and someone here is the Debian project. Squeeze released with kfreebsd packages as part of the main archive, which is again more than sarge did for amd64. Yes, they're labelled as technology previews, but that doesn't make them unofficial, nor not part of the main archive. I'm talking about what kernel people use when they use Debian, and what platform development that creates the software in the distribution happens on. There's no plan to migrate from Linux to BSD kernel That's somewhat of a straw man. No-one's ever suggested that support for Linux be dropped, so far as I'm aware. There's no reason why we can't at least try our best to support multiple kernels, so there's no migration to plan or not plan for (and no, systemd only supports Linux is not an argument against supporting other kernels). Regards, Adam -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/130099.8028.26.ca...@hathi.jungle.funky-badger.org
Re: [Lennart Poettering] Re: A few observations about systemd
Wouter Verhelst wouter at debian.org writes: Debian is the 'Universal' operating system, and many of our developers (including myself) pride themselves on that. We port to many architectures, we port to multiple kernels. It's one of the defining features of Debian: you can run it /anywhere/ This is an almost religious argument. You take the value of running on multiple kernels as an article of faith, with no evaluation of the benefits (either to people directly using such ports, or possible feedback to the main distribution). It's hard to make rational arguments against such a position, other than to note that the position is irrational and causes practical harm where it interferes with rational decisions. I think one fallacy underlying that position is to consider supporting a new first-class object such as a new kernel as a big step towards universality even if it isn't actually that useful in many cases - being usable in one more niche Linux subcase could well mean more in practice than being usable with BSD kernel. I wonder how many developers actually pride themselves in the existence of the kFreeBSD or Hurd ports. I hope not too many. The Debian Linux distribution does have real value for users and for the development of free software. -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/loom.20110719t230956-...@post.gmane.org
Re: [Lennart Poettering] Re: A few observations about systemd
Wouter Verhelst wrote: Debian is the 'Universal' operating system Universal operating system is a phrase that was added to our website most likely by Bruce Perens in April 1997. (He posted about it to -private then; there was no discussion.) Not very coincidentally, Debian made 3 press releases around that time, about the space shuttle flights. So beyond a certian resonance, this PR phrase has about as much modern significance as Vice President of Engineering, which is another phrase Bruce inflicted in Debian around the same time. If you're looking for a general-purpose statement of what Debian is that is vague enough to use to support any argument[1], universal operating system should be considered deprecated; instead look to the social contract. -- see shy jo [1] FWIW, the first use of universal operating system to back up a random POV seems to have been used in an argument about keeping the purity test out of Debian. That argument failed. signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
On Tue, Jul 19, 2011 at 03:59:13PM +, Uoti Urpala wrote: Wouter Verhelst wouter at debian.org writes: IMAO, a statement of (paraphrased) 'portability is for weenies' isn't Keeping portability in mind is a good thing especially if you're doing something that is easily implementable with common interfaces. However in some cases additional portability has very real costs, and it's by no means given that the balance should go in the favor of portability. … and neither should the balance go in favour of dropping portability. In my experience, programs written with portability in mind are much more resilient to breakage, and thus over time they survive bit-rot much better. Whenever I see a program that is explicitly non-portable, I tend to discount it in favour of portable alternatives, because it means: - the author has considered multiple alternatives, and doesn't rely on Linux kernel x.y.z especially or glibc version n - if a given feature is deprecated, the program might still work by falling back to another feature (re. epoll-vs-poll), possibly in a degraded mode but still work - any many other considerations So, while you have said very clearly in this thread portability should be amongst the last considerations, understand that not everyone shares your point of view. regards, iustin signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
Adam D. Barratt adam at adam-barratt.org.uk writes: Proof by assertion isn't an argument. If you think kfreebsd sucks then you're entitled to that opinion, but please don't seek to frame it as some sort of consensus direction on the part of the project because it's obvious. What I consider obvious is not direct agreement on kFreeBSD sucks, but that it will not become a significant platform for Debian use and development. Do you seriously expect to see major popularity for it? I know there have been official Debian/kFreeBSD releases, but those make very little difference - someone published a set of files which were then ignored by about everyone. fwiw, they're more official than the original amd64 release of Debian Yes, more _official_. Doesn't that support what I said? AMD64 was not official, while kFreeBSD was. Shows that the official label is not all that relevant for practical considerations. I'm talking about what kernel people use when they use Debian, and what platform development that creates the software in the distribution happens on. There's no plan to migrate from Linux to BSD kernel That's somewhat of a straw man. No-one's ever suggested that support for Linux be dropped, so far as I'm aware. There's no reason why we That no-one's ever suggested that support for Linux be dropped is close to the point I was trying to make. Nobody seriously thought kFreeBSD would become the center of Debian use/development, much less suggested deprecating Linux. kFreeBSD was accepted not as a future direction of Debian but as one more niche thing in the archive. can't at least try our best to support multiple kernels, so there's no migration to plan or not plan for (and no, systemd only supports Linux is not an argument against supporting other kernels). I've never argued directly against support for other kernels (if you thought so you've misunderstood something). What I've said is that decisions for Linux should be made without undue concern for niche ports. If systemd is good _for Linux_ then systemd should be used, and the burden for any compatibility work should be on the people who want to maintain the other ports. -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/loom.20110719t233208-...@post.gmane.org
Re: [Lennart Poettering] Re: A few observations about systemd
Iustin Pop iustin at debian.org writes: In my experience, programs written with portability in mind are much more resilient to breakage, and thus over time they survive bit-rot much better. Whenever I see a program that is explicitly non-portable, I tend to discount it in favour of portable alternatives, because it means: I believe it's true that over some general sample of programs portable ones have higher quality. However the main reason for that is that skilled programmers are more likely to be aware of portability concerns and not just try changing code until the find the first version that runs on their machine. If you have a programmer of given skill then I doubt whether he tries to create a portable program or not has all that much effect on the quality. I think there is more and better evidence available to evaluate the quality of systemd code and the skill of its authors than correlation with portability (and I'm pretty sure they'd have the skill to write portable code). So, while you have said very clearly in this thread portability should be amongst the last considerations, understand that not everyone shares your point of view. I don't think that accurately represents my view. Most of the C code I've written has been quite portable (standard-conforming and using POSIX functions only, though in most cases without trying to work around any flaws of specific compilers). I just don't think portability would be of such overriding concern that it would be THE attribute to evaluate code quality by, or that going for maximum portability would be the universal right answer in every case. -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/loom.20110720t000835-...@post.gmane.org
Re: [Lennart Poettering] Re: A few observations about systemd
On Tue, Jul 19, 2011 at 09:52:30PM +, Uoti Urpala wrote: Adam D. Barratt adam at adam-barratt.org.uk writes: Proof by assertion isn't an argument. If you think kfreebsd sucks then you're entitled to that opinion, but please don't seek to frame it as some sort of consensus direction on the part of the project because it's obvious. What I consider obvious is not direct agreement on kFreeBSD sucks, but that it will not become a significant platform for Debian use and development. Do you seriously expect to see major popularity for it? Yes. Your point? can't at least try our best to support multiple kernels, so there's no migration to plan or not plan for (and no, systemd only supports Linux is not an argument against supporting other kernels). I've never argued directly against support for other kernels (if you thought so you've misunderstood something). What I've said is that decisions for Linux should be made without undue concern for niche ports. You clearly misunderstand how Debian works. Anyway, you're mostly trolling (and according to google, that seems to be your regular M.O.), so EOT for me. -- The volume of a pizza of thickness a and radius z can be described by the following formula: pi zz a -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110719222752.ge6...@grep.be
Re: [Lennart Poettering] Re: A few observations about systemd
[Joey Hess] Universal operating system is a phrase that was added to our website most likely by Bruce Perens in April 1997. (He posted about it to -private then; there was no discussion.) Huh. I always assumed Bdale coined it for his DPL platform. Turns out this thread wasn't a _total_ waste of time after all. -- Peter Samuelson | org-tld!p12n!peter | http://p12n.org/ -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110719224237.gh4...@p12n.org
Re: [Lennart Poettering] Re: A few observations about systemd
On Tue, Jul 19, 2011 at 05:31:37PM -0400, Joey Hess wrote: Wouter Verhelst wrote: Debian is the 'Universal' operating system Universal operating system is a phrase that was added to our website most likely by Bruce Perens in April 1997. (He posted about it to -private then; there was no discussion.) Not very coincidentally, Debian made 3 press releases around that time, about the space shuttle flights. So beyond a certian resonance, this PR phrase has about as much modern significance as Vice President of Engineering, which is another phrase Bruce inflicted in Debian around the same time. The key difference being that the former stuck, while the latter didn't. Where it came from is less important than what it represents today to some of us. I believe I can read in your post that you don't like it; but certainly this is not true for all of us. -- The volume of a pizza of thickness a and radius z can be described by the following formula: pi zz a -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110719225236.gk15...@grep.be
Re: [Lennart Poettering] Re: A few observations about systemd
m...@linux.it (Marco d'Itri) writes: On Jul 19, Wouter Verhelst wou...@debian.org wrote: kFreeBSD is currently released as a technology preview. With that, we mean it works, but it isn't necessarily ready yet for prime usage. The fact that there currently aren't many users yet isn't surprising in that light. However, as the port matures, it's not unthinkable that there will be many more users. Again, why? ZFS is a pretty big one. -- Russ Allbery (r...@debian.org) http://www.eyrie.org/~eagle/ -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/878vrtpxxs@windlord.stanford.edu
Re: [Lennart Poettering] Re: A few observations about systemd
On Sun, Jul 17, 2011 at 06:51:17PM +0200, Lennart Poettering wrote: In fact, a minimal systemd system will win in almost very aspect against a remotely similarly powerful sysvinit system: you will need much fewer processes to boot. That means much shorter boot times. This is, as far as I'm aware, an unproven assertion. While it's true that there is a cost to the additional processes used in init scripts, I have not seen any serious attempt at measuring how big this impact actually is - certainly not in terms that would be relevant to Debian, which uses dash as its /bin/sh and insserv by default (in squeeze and above). I'm sure that systemd does much better than a traditional sysvinit boot with /bin/bash and no dependency-based booting. But then, so does Debian's current boot system, and so does upstart; and neither of the latter two involve grandiose claims of a shell-free boot. Trying to take the shell completely out of the boot means a definite tradeoff here between boot speed and configurability/maintainability, and in the absence of hard numbers, I suspect this is a false optimization and not a trade-off that we actually want to make in a general distribution. Which then calls into question the use of such claims as a justification for a switch to systemd at all... For a low-level piece of infrastructure, systemd interacts with a lot of high-level software; while this might be okay for a workstation running Gnome, it makes me wonder whether it will be usable on servers. A cursory look shows that systemd intrinsically depends on D-Bus (the *desktop* bus) and knows about Plymouth, RedHat's implementation of a splash screen. More on that below. Oh, come on. systemd does not depend on Plymouth, it merely interacts with it if it is around it. Where interaction simply means writing a single message every now and then to ply to keep it updated how far the boot proceeded. It's more or a less a single line of text we send over every now and then in very terse code. One of these days I'll get around to writing that blog entry to set the record straight on why plymouth is an indispensible component of boot with any modern boot system, because when everything is starting in parallel, you need something to handle I/O multiplexing to the user on console. So in a real sense, it *should* be a dependency. Even if you don't care about splash, you still need multiplexing. Upstart has the same dependency, though pid 1 doesn't talk directly to plymouth in the upstart model; instead this is handled by an out-of-process plymouth-upstart bridge, as well as by the out-of-process mountall service that talks to plymouth for handling of fscks, skipping the mounting of missing filesystems, etc. He practices misleading advertising[2], likes to claim that the universal adoption of systemd by all distributions is a done thing[3], and attempts to bully anyone who has the gall to think that the discussion is still open[4]. Juliusz practices misleading anti-advertising [1], likes to ignore the fact that all major distros either made systemd the default or include it in their distro with the exception of Ubuntu. Well, it's nice to see that Lennart is at least acknowledging Ubuntu as a major distribution these days, unlike in some of his earlier rhetoric. ;) Though this is still a pretty misleading comment, since both of these statements are also true: All major distros either made sysvinit the default or include it in their distro. All major distros either made upstart the default or include it in their distro. Ho-hum... -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. Ubuntu Developerhttp://www.debian.org/ slanga...@ubuntu.com vor...@debian.org signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
On Mon, Jul 18, 2011 at 01:22:37PM -0700, Steve Langasek wrote: On Sun, Jul 17, 2011 at 06:51:17PM +0200, Lennart Poettering wrote: In fact, a minimal systemd system will win in almost very aspect against a remotely similarly powerful sysvinit system: you will need much fewer processes to boot. That means much shorter boot times. This is, as far as I'm aware, an unproven assertion. While it's true that there is a cost to the additional processes used in init scripts, I have not seen any serious attempt at measuring how big this impact actually is - certainly not in terms that would be relevant to Debian, which uses dash as its /bin/sh and insserv by default (in squeeze and above). I'm sure that systemd does much better than a traditional sysvinit boot with /bin/bash and no dependency-based booting. But then, so does Debian's current boot system, and so does upstart; and neither of the latter two involve grandiose claims of a shell-free boot. Trying to take the shell completely out of the boot means a definite tradeoff here between boot speed and configurability/maintainability, and in the absence of hard numbers, I suspect this is a false optimization and not a trade-off that we actually want to make in a general distribution. Which then calls into question the use of such claims as a justification for a switch to systemd at all... I'd expect some important differences between shell script based init and systemd-type init by the simple fact that there are (or at least should be, I don't know how systemd actually works) less files to read. Less files to read == less disk seeks. Disks seeks hurt startup performance. Mike -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110718211439.ga14...@glandium.org
Re: [Lennart Poettering] Re: A few observations about systemd
Steve Langasek vorlon at debian.org writes: I'm sure that systemd does much better than a traditional sysvinit boot with /bin/bash and no dependency-based booting. But then, so does Debian's current boot system, and so does upstart; and neither of the latter two involve grandiose claims of a shell-free boot. Trying to take the shell completely out of the boot means a definite tradeoff here between boot speed and configurability/maintainability, and in the absence of hard numbers, I Tradeoff? What tradeoff? Sysv-style init scripts are messy, definitely not maintainable, and theoretically configurable in the turing-complete sense but hard to modify in practice. systemd service configuration wins in boot speed, wins big in maintainability, and is at least equal in configurability (you can configure most things easier than with shell scripts, but can still fall back to them if necessary). Juliusz practices misleading anti-advertising [1], likes to ignore the fact that all major distros either made systemd the default or include it in their distro with the exception of Ubuntu. Well, it's nice to see that Lennart is at least acknowledging Ubuntu as a major distribution these days, unlike in some of his earlier rhetoric. ;) Though this is still a pretty misleading comment, since both of these statements are also true: All major distros either made sysvinit the default or include it in their distro. All major distros either made upstart the default or include it in their distro. It's not that misleading after all when you consider how quickly systemd has reached that position. It was only published last year. To have reached its current position already shows a lot of momentum. Sysvinit is the old default, but nobody seriously claims it's gaining ground. Upstart is still used in Ubuntu but doesn't seem to have much future elsewhere. There's quite a lot of interest in systemd for Debian too, whereas I've seen few people express interest in Upstart. The interest is especially low when you consider Debian's ties with Ubuntu and people who only care about Upstart because of that. -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/loom.20110719t001733-...@post.gmane.org
Re: [Lennart Poettering] Re: A few observations about systemd
Uoti Urpala uoti.urp...@pp1.inet.fi writes: Upstart is still used in Ubuntu but doesn't seem to have much future elsewhere. There's quite a lot of interest in systemd for Debian too, whereas I've seen few people express interest in Upstart. Funny, my personal experience has been the exact opposite, including the conversations that I had in-person at the last Debconf. The interest is especially low when you consider Debian's ties with Ubuntu and people who only care about Upstart because of that. This is completely false. I have no affiliation whatsoever with Ubuntu and personally have no interest in using it. Nonetheless, I think upstart looks quite interesting systemd also looks interesting, and I'm generally happy with either of them as possibilities, I'm rather concerned by systemd's lack of interest in portability. The upstart maintainers have expressed considerably more willingness to date to work with Debian on meeting our project's goals and incorporating those changes into the upstream release. -- Russ Allbery (r...@debian.org) http://www.eyrie.org/~eagle/ -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87hb6jqkx6@windlord.stanford.edu
Re: [Lennart Poettering] Re: A few observations about systemd
Russ Allbery r...@debian.org (18/07/2011): The upstart maintainers have expressed considerably more willingness to date to work with Debian on meeting our project's goals and incorporating those changes into the upstream release. For reference, that would likely be: http://lists.debian.org/debian-bsd/2009/07/msg00122.html (Feel free to correct me if you had other references in mind.) Mraw, KiBi. signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
On Mon, Jul 18, 2011 at 10:18:14PM +, Uoti Urpala wrote: Steve Langasek vorlon at debian.org writes: I'm sure that systemd does much better than a traditional sysvinit boot with /bin/bash and no dependency-based booting. But then, so does Debian's current boot system, and so does upstart; and neither of the latter two involve grandiose claims of a shell-free boot. Trying to take the shell completely out of the boot means a definite tradeoff here between boot speed and configurability/maintainability, and in the absence of hard numbers, I Tradeoff? What tradeoff? The tradeoff of hard-coding policy into C code in exchange for faster boot. Sysv-style init scripts are messy, definitely not maintainable, and theoretically configurable in the turing-complete sense but hard to modify in practice. Yes, all of this is true. You seem to have mistaken my criticism of the systemd model for a defense of sysvinit, which it was not. A system where *everything* is a shell script is not very maintainable; but neither is a system whose design is predicated on the idea that everything is built-in. The middle ground between the two seems to be upstart. systemd service configuration wins in boot speed, You did actually read my message, right, where I observed that there are no published numbers to support this claim in a relevant head-to-head comparison? And your only response is to repeat the unsubstantiated claim? Though this is still a pretty misleading comment, since both of these statements are also true: All major distros either made sysvinit the default or include it in their distro. All major distros either made upstart the default or include it in their distro. It's not that misleading after all when you consider how quickly systemd has reached that position. Um, of course it is. Claiming that it's included in the distro as if that's some sort of major milestone is *incredibly* misleading. Lots of things are included in lots of distros that are never going to be used by default. Debian has certainly not made a decision to use systemd yet, but that doesn't stop Lennart from using the package's presence in the Debian archive in his propaganda. -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. Ubuntu Developerhttp://www.debian.org/ slanga...@ubuntu.com vor...@debian.org signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
Cyril Brulebois k...@debian.org writes: Russ Allbery r...@debian.org (18/07/2011): The upstart maintainers have expressed considerably more willingness to date to work with Debian on meeting our project's goals and incorporating those changes into the upstream release. For reference, that would likely be: http://lists.debian.org/debian-bsd/2009/07/msg00122.html (Feel free to correct me if you had other references in mind.) I think this is somewhat obsoleted by in-person discussions at Debconf in 2010. But this is now-fallible year-old memory, so having the discussion with him explicitly to get the current state of his thinking would be a good idea. We talked about BSD explicitly during the various init system discussions at Debconf, and the impression I came away with was that he didn't have the time to write the code himself, but was definitely willing to work with someone who was interested and make sure that it would continue to work. -- Russ Allbery (r...@debian.org) http://www.eyrie.org/~eagle/ -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/871uxnqk5z@windlord.stanford.edu
Re: [Lennart Poettering] Re: A few observations about systemd
I have not seen any serious attempt at measuring how big this impact actually is I'd expect some important differences between shell script based init and systemd-type init Yeah, that's everybody's intuition too. But Steve is right -- it would be good to see some real benchmarks. -- Juliusz -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87sjq3fc2k@trurl.pps.jussieu.fr
Re: [Lennart Poettering] Re: A few observations about systemd
On Mon, Jul 18, 2011 at 11:14:39PM +0200, Mike Hommey wrote: I'm sure that systemd does much better than a traditional sysvinit boot with /bin/bash and no dependency-based booting. But then, so does Debian's current boot system, and so does upstart; and neither of the latter two involve grandiose claims of a shell-free boot. Trying to take the shell completely out of the boot means a definite tradeoff here between boot speed and configurability/maintainability, and in the absence of hard numbers, I suspect this is a false optimization and not a trade-off that we actually want to make in a general distribution. Which then calls into question the use of such claims as a justification for a switch to systemd at all... I'd expect some important differences between shell script based init and systemd-type init by the simple fact that there are (or at least should be, I don't know how systemd actually works) less files to read. Less files to read == less disk seeks. Disks seeks hurt startup performance. Yes, I've read your blog entries on the subject. :-) That's true, but I think the reduction in the number of files being accessed, for systemd vs. sysvinit or upstart, is rather small; aside from some things in /etc/rcS.d, most init scripts would have approximately a 1:1 correlation with upstart jobs or systemd config files, and if you've read the shell off disk once it's in cache and there's not likely to be any more seeking. So I do expect that most of the shell penalty will be CPU rather than disk in the context of boot. -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. Ubuntu Developerhttp://www.debian.org/ slanga...@ubuntu.com vor...@debian.org signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
Russ Allbery rra at debian.org writes: Uoti Urpala uoti.urpala at pp1.inet.fi writes: Upstart is still used in Ubuntu but doesn't seem to have much future elsewhere. There's quite a lot of interest in systemd for Debian too, whereas I've seen few people express interest in Upstart. Funny, my personal experience has been the exact opposite, including the conversations that I had in-person at the last Debconf. Last? As in 2010? Given how quickly systemd has gained momentum, the 2010 status is hardly representative of current interest in systemd or its relative popularity compared to Upstart. The interest is especially low when you consider Debian's ties with Ubuntu and people who only care about Upstart because of that. This is completely false. I have no affiliation whatsoever with Ubuntu and personally have no interest in using it. Nonetheless, I think upstart looks quite interesting I didn't claim that there would not be a single person interested in Upstart. You could be more familiar with the attitudes of Debian developers than I am, but if 2010 experience and your personal opinion are the only things you're basing that on then it's not enough to show that low interest in Upstart would be completely false (even restricted to developers only). systemd also looks interesting, and I'm generally happy with either of them as possibilities, I'm rather concerned by systemd's lack of interest in portability. The upstart maintainers have expressed considerably more willingness to date to work with Debian on meeting our project's goals and incorporating those changes into the upstream release. I think the important question is whether portability to other kernels is or should be a project's goal, and how much else you're willing to lose for the sake of that goal. I know I would personally be a lot happier with a Debian that supports systemd functionality than with a Debian that can run on a BSD kernel. -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/loom.20110719t004726-...@post.gmane.org
Re: [Lennart Poettering] Re: A few observations about systemd
Uoti Urpala uoti.urp...@pp1.inet.fi writes: I think the important question is whether portability to other kernels is or should be a project's goal, and how much else you're willing to lose for the sake of that goal. I believe that it should be, and I'm willing to lose systemd for that goal, although hopefully it wouldn't come to that. I know I would personally be a lot happier with a Debian that supports systemd functionality than with a Debian that can run on a BSD kernel. Well, while we're putting stakes in the ground, I suppose I'll hammer mine in there as well. I completely disagree to the point that I would take that to a GR. Thankfully, I suspect this will be a false dichotomy. -- Russ Allbery (r...@debian.org) http://www.eyrie.org/~eagle/ -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/874o2jp3fm@windlord.stanford.edu
Re: [Lennart Poettering] Re: A few observations about systemd
On Mon, Jul 18, 2011 at 11:05:56PM +, Uoti Urpala wrote: I think the important question is whether portability to other kernels is or should be a project's goal, and how much else you're willing to lose for the sake of that goal. I know I would personally be a lot happier with a Debian that supports systemd functionality than with a Debian that can run on a BSD kernel. I ran GNU/kFreeBSD on a server of mine for over a year because it had pf. pf makes OS fingerprinting automatic and a lot easier (at the time, Debian's Linux kernel did not have the osf module) and traffic shaping is much, much easier as well[0]. The Linux kernel has only recently had ipset functionality merged upstream, which pf has had for years. The FreeBSD kernel also had a much, much more responsive scheduler as well (it may still, I don't know). It also supports ZFS, which is very important to some people. The reason I left is because pf stopped working. I agree that GNU/kFreeBSD is not a great desktop platform, but it is an excellent server platform. Also, I've installed systemd on my laptop and it logs almost nothing to the console (verbose on the kernel command line does not help). Logging to syslog is not helpful when the system won't come up to the point of starting syslog. What it *does* log (to syslog), however, is a message that /usr as a separate partition is obsolete, even though this has no effect on systemd at all, other than offending the upstream author. Last I checked, The Unix Way did not involve having important system programs prattle on about irrelevant details. I'll side with supporting GNU/kFreeBSD over systemd any day. [0] Extremely limited bandwidth for incoming Windows SMTP servers, anyone? -- brian m. carlson / brian with sandals: Houston, Texas, US +1 832 623 2791 | http://www.crustytoothpaste.net/~bmc | My opinion only OpenPGP: RSA v4 4096b: 88AC E9B2 9196 305B A994 7552 F1BA 225C 0223 B187 signature.asc Description: Digital signature
Re: [Lennart Poettering] Re: A few observations about systemd
Steve Langasek vorlon at debian.org writes: Tradeoff? What tradeoff? The tradeoff of hard-coding policy into C code in exchange for faster boot. What's actually hard-coded so hard that it would have negative effects? What do you actually *lose* here? The systemd model prefers to avoid shell scripts when possible. I think that's a very good principle. But it's not like you couldn't run shell code if you can't achieve the effect you want any other way (after all even compatibility for sysv scripts is still provided!). By the way, I think in exchange for faster boot is focusing too narrowly on boot speed. It's not like boot speed would be the only reason to avoid shell. Sysv-style init scripts are messy, definitely not maintainable, and theoretically configurable in the turing-complete sense but hard to modify in practice. Yes, all of this is true. You seem to have mistaken my criticism of the systemd model for a defense of sysvinit, which it was not. A system where *everything* is a shell script is not very maintainable; but neither is a system whose design is predicated on the idea that everything is built-in. The middle ground between the two seems to be upstart. Again, what's the actual maintainability problem in the systemd model? I think you haven't identified any, and I can't really guess what you mean either. systemd service configuration wins in boot speed, You did actually read my message, right, where I observed that there are no published numbers to support this claim in a relevant head-to-head comparison? And your only response is to repeat the unsubstantiated claim? I don't know how much it wins, and I don't really care that much about boot speed myself. Also, overall speed win could come from socket activation too, not just avoidance of shell scripts. My main point was that your claim of tradeoff was unsubstantiated as you didn't actually identify any negative effects to counter the speed gains (and in fact I think quite the opposite is true). That holds whether the speed gains are large or small. It's not that misleading after all when you consider how quickly systemd has reached that position. Um, of course it is. Claiming that it's included in the distro as if that's some sort of major milestone is *incredibly* misleading. Lots of things are included in lots of distros that are never going to be used by default. Debian has certainly not made a decision to use systemd yet, but that doesn't stop Lennart from using the package's presence in the Debian archive in his propaganda. Yes literally just is included doesn't mean much, but it has more support in Debian than just a lone maintainer uploading it. Anyway I don't want to argue about the exact semantics of his statement. systemd does have a lot momentum even if its adoption is not a done deal in every distribution yet, and it's hard to imagine Upstart turning the tide now or a new candidate appearing and taking over. -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/loom.20110719t011655-...@post.gmane.org
Re: [Lennart Poettering] Re: A few observations about systemd
]] brian m. carlson Hi, | Also, I've installed systemd on my laptop and it logs almost nothing | to the console (verbose on the kernel command line does not help). try doing systemd.log_level=debug as documented in the man page? cheers, -- Tollef Fog Heen UNIX is user friendly, it's just picky about who its friends are -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87bowqzvhs@qurzaw.varnish-software.com