Mailinglist Archive: opensuse (982 mails)

< Previous Next >
Re: [opensuse] Getting rid of systemd and putting sysv back
  • From: Timothy Butterworth <timothy.m.butterworth@xxxxxxxxx>
  • Date: Wed, 24 Sep 2014 11:10:33 -0500
  • Message-id: <CAO6YxPzq_0L0RXsX_DdLbH7i9Vjeaj=Em=kBqfQBvxP45QAqOw@mail.gmail.com>
Some of The BSD Developers have already been talking about porting
systemd over to BSD. With Debian, Red Hat Inc and SUSE migrating to
systemd pretty much all cloned GNU/Linux distros are migrating.

I like a lot of what systemd is doing but there are some things I do
not like. journald is one of my concerns placing logs into a binary
format may be useful to help prevent compromised systems from having
logging entries removed. With text files based logging it is
relatively easy to remove some lines from the log to cover up the
system compromise. Of course using an external log collector and
forwarding copies of all entries to it helps to eliminate this threat.
SystemD does allow for RSyslog or SyslogNG which good as plain text
logs do have the added advantage of easy of use with a massive amount
of GNU utilities to parse the with. I currently am unclear as to
whether journald can even send copies to a remote log collection
server but I am sure the developers have thought of this.

Security vs ease of use is always a sticky situation. The kernel
cgroups tracking feature is a great security enhancement and stability
enhancement so intentionally malicious multi forked PID's can not be
made hidden and used to prevent malicious process from being stopped
or restarted as part of the parent process using sytemvint.

I do not like the HTTP server integration at all and I do not want
this at all journald is optional though.

Gentoo still has not embraced systemd but it is an implemented option
and required for GNOME 3. As long as you do not want to use GNOME 3
(Impossible now without systemd now) you can use Gentoo with a
different choice of DE and systemvinit.

There is at least a commercial book available, Fedora Linux Servers
with systemd, that covers the Fedora 20 implementation of systemd.
Once you understand the syntax it is at least easy to use and
consistent for an OSS moving target development project. Sadly the
authors writing style is that of a Historian and not a technology
enthusiast though.

I really do not care for the idea of integrating DBus into systemd and
PID 1. I have heard some of the concerns being voiced about this and
it sounds like trouble waiting to happen.

Bottom line in the long run I personally think systemd will bring a
lot to Linux even if it does require being forked sometime in the
future to overhaul some aspects of it so its development is not a
waist at all in my opinion. Once the larger community steps in and
possibly forks it and host the fork at The Linux Foundation if
possible I imagine a lot of these debates will end with more folks
actually working on it. The beauty of FLOSS if you do not like a
direction something is taking and your concerns are not adhered to
then stick a fork in it and prove yours is better.

If systemd was a Linux Foundation working group then LSB and other
requirements could be added into this umbrella to create some useful
standardization. One day their will be a Universal Linux Package
format and package manager after someone actually convinces all the
developers of current solutions to work together to produce a new a
new one from scratch!

Some of the arguments against systemd though seem to be being spoon
fed by a GNU/Linux Hater or FUD campaign because they actually attack
both systemvinit and systemd.
--
To unsubscribe, e-mail: opensuse+unsubscribe@xxxxxxxxxxxx
To contact the owner, e-mail: opensuse+owner@xxxxxxxxxxxx

< Previous Next >
This Thread
Follow Ups