On Jan 24, 2016, at 6:40 PM, Peter Duffy <peter at pwduffy.org.uk> wrote:> The thing which always gets me about systemd is not the thing itself, > but the way it was rolled out. When I first installed Red Hat 7, if a > window had appeared telling me about systemd and asking me if I wanted > to use it, or stick with the old init framework, I'd have opted for the > latter (as I was interested primarily in continuity from the previous > version.) But I'd have noted the existence of systemd, and would have > tried it out on a sacrificial box - I might even have got to like it! > But having it rammed down my throat just put me off it for life (bit > like a kid being force-fed Brussels sprouts.)It wasn?t a huge surprise. systemd was in Fedora since f15, and RHEL7 was branched from f18 (iirc). systemd was in the RHEL7.0 beta. The release announcement was filled with information about systemd.[1] Frankly, I was more surprised about XFS and 64-bit-only than systemd. I believe that RHEL7 (and CentOS7) both have systemd integrated into them enough that it isn?t as simple as ?choose init system? on install. Whether you like it or not, systemd has its fingers in a lot of stuff, like login services, resource management, stuff like tmpfile creation and management. I?m not exactly thrilled with some features (like the way systemd ?user was implemented, the inflexibility of cgroup configuration per-unit, remote journal forwarding) but overall, since I was prepared, I think its a step in the right direction. If there was one thing I?d love, is for there to be a systemd long term support release. I feel like the systemd in el7.0 was way too early, and it wasn?t until 7.2 that I feel like things are starting to stabilize. Also, thankfully, systemd ?user was *completely torn out* in 7.2. :) As for Gnome3, I simply don?t use it. lightdm + cinnamon or MATE for me. 1. https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/7/html/7.0_Release_Notes/pref-Red_Hat_Enterprise_Linux-7.0_Release_Notes-Introduction.html ? Jonathan Billings <billings at negate.org>
> I believe that RHEL7 (and CentOS7) both have systemd integrated into them enough > that it isn?t as simple as ?choose init system? on install.That's because of systemd. Even if most of the linux distros don't, giving the choice is a bit less difficult with any other init system. The main problem is systemd makes (often badly) more and more things that, as a "simple" init system, it should not do (login, and "su-ing" now, journaling, device management via udev, and so on), violating the KISS principle. If you use systemd, you have to use all the systemd tentacles, even if you don't want. Worse, more and more programs hardly depend on systemd now. Gnome 3 is an example, and that's why I don't, and won't, use Gnome 3. I don't use any systemd-based distro personally. Sadly, professionally, I have to, since RedHat/CentOS and Debian adopted it (and Ubuntu LTS will do soon). And systemd makes my job uselessly more complicated. For exemple, why must I deal with journald and its fancies when I setup a syslog server (and I have to, because journald don't even know what are centralized logs...) on my servers ? Why systemd maintainers continuously change big parts of its behaviour, without any consideration of major-minor versionning, and why RedHat/CentOS maintainers dismiss this fact (the CentOS 7.1 to 7.2 update is painful, because systemd switched from 208 to 219) ? Why, more generally, the answer is often "systemd" when I encounter a problem on a server ? Sylvain. Pensez ENVIRONNEMENT : n'imprimer que si ncessaire
On Mon, January 25, 2016 7:59 am, Sylvain CANOINE wrote:> >> I believe that RHEL7 (and CentOS7) both have systemd integrated into >> them enough >> that it isn???t as simple as ???choose init system??? on install. > > That's because of systemd. Even if most of the linux distros don't, giving > the choice is a bit less difficult with any other init system. > > The main problem is systemd makes (often badly) more and more things that, > as a "simple" init system, it should not do (login, and "su-ing" now, > journaling, device management via udev, and so on), violating the KISS > principle. If you use systemd, you have to use all the systemd tentacles, > even if you don't want. Worse, more and more programs hardly depend on > systemd now. Gnome 3 is an example, and that's why I don't, and won't, use > Gnome 3. > > I don't use any systemd-based distro personally. Sadly, professionally, I > have to, since RedHat/CentOS and Debian adopted it (and Ubuntu LTS will do > soon). And systemd makes my job uselessly more complicated. For exemple, > why must I deal with journald and its fancies when I setup a syslog server > (and I have to, because journald don't even know what are centralized > logs...) on my servers ? Why systemd maintainers continuously change big > parts of its behaviour, without any consideration of major-minor > versionning, and why RedHat/CentOS maintainers dismiss this fact (the > CentOS 7.1 to 7.2 update is painful, because systemd switched from 208 to > 219) ? Why, more generally, the answer is often "systemd" when I encounter > a problem on a server ?If that is the case, why do you run CentOS 7 on the server? You can stay with CentOS 6 for now and either wait till Linux systemd-free distribution mature enough to be run on server is available. Which it almost is: Devuan (systemd-free fork of Debian) has released "alpha" version about half a year ago. If you feel "married" to Linux, maybe it is a good idea to play with Devuan, provide them feedback thus helping them to become system-free Linux acceptable for servers. Simultaneously you can explore other options which would be to migrate away from Linux (Open Solaris, FreeBSD, NetBSD, OpenBSD will be much smaller step than stepping up to CentOS 7 - that is my experience, though FreeBSD migration of servers I started came much earlier than CentOS 7 and for different reason). Good Luck! Valeri> > Sylvain. > Pensez ENVIRONNEMENT : n'imprimer que si ncessaire >+++++++++++++++++++++++++++++++++++++++ Valeri Galtsev Sr System Administrator Department of Astronomy and Astrophysics Kavli Institute for Cosmological Physics University of Chicago Phone: 773-702-4247 ++++++++++++++++++++++++++++++++++++++++