systemd or why will user fall away from fedora?

[Farkas Levente]

hi,
why i not like the idea of systemd?
it's something that dramatically change the system behavior. period.
this is different from all other/previous unix/linux system.
there're many config scripts, programs and third party packages which
are assume the old setup.
such basic changes need years for adaptation and testing.
and we all know what happened with pulseaudio. everybody turn it off and
remove it from the system in at least 3 fedora release, just because
some of their apps are not working.
imho it's the right time to release rhel-6 which still use upstart and
no one have to use systemd for a few years.
for real servers sysadms usually not use fedora. but i'm sure after that
move no one will use fedora for servers.
for workstation most users already use ubuntu. why? because it's more
user friendly.
do you think workstation users will like this kind of changes?
do you think it'll be good to diverge from all other linux distros?
do you think workstation users will see any kind of uplift because of
systemd?

imho none of the technical questions are real questions, these are the
real questions.

-- 
  Levente                               "Si vis pacem para bellum!"

[Till Maas]

On Tue, Aug 24, 2010 at 10:46:41PM +0200, Farkas Levente wrote:

>  for workstation most users already use ubuntu. why? because it's
> more
>  user friendly. 
There is nothing wrong with using Ubuntu, if it servers their needs.

>  do you think workstation users will like this kind of changes?
If they like a faster bootup, then yes, they will. And I as a
workstation user like it.

>  do you think it'll be good to diverge from all other linux
> distros? 
As long as we diverge in the direction of progress, then yes, it is. It
is a core principle of Fedora. Also this diversion is the only reason
why there are several distributions. The change might even drive more
users to Fedora.

>  do you think workstation users will see any kind of uplift because
> of
>  systemd? 
Yes, they get a faster bootup afaik and happier package maintainers. ;-)

Regards
Till

[Mike McGrath]

On Tue, 24 Aug 2010, Till Maas wrote:

>  On Tue, Aug 24, 2010 at 10:46:41PM +0200, Farkas Levente wrote:
>
>  > for workstation most users already use ubuntu. why? because it's more
>  > user friendly.
>
>  There is nothing wrong with using Ubuntu, if it servers their needs.
>
>  > do you think workstation users will like this kind of changes?
>
>  If they like a faster bootup, then yes, they will. And I as a
>  workstation user like it.
>  
[citation needed]

I asked for this and was told by developers they were reluctant to post
the data.

	-Mike

[Christof Damian]

On Tue, Aug 24, 2010 at 23:06, Mike McGrath <mmcgrath(a)redhat.com&gt; wrote:
>  On Tue, 24 Aug 2010, Till Maas wrote:
> > If they like a faster bootup, then yes, they will. And I as a
> > workstation user like it.
> >
>
>  [citation needed]
>
>  I asked for this and was told by developers they were reluctant to post
>  the data. 
I also wonder who these people are who boot their computers so often
that boot-up speed makes any difference over a whole day.

I like the on-demand service features of systemd though. I just hate
that Fedora will be a different beast from any other Linux for the
foreseeable future.

It will also be interesting to see what this will mean in relation to
RHEL, because one of the reasons I use Fedora is that I can transfer
stuff I learn and use on Fedora to RHEL/CentOS. If they are very
different I can as well choose one SuSe (if I want to stick with RPM
at least).

Christof

[Nathaniel McCallum]

On 08/25/2010 06:20 AM, Christof Damian wrote:
>  On Tue, Aug 24, 2010 at 23:06, Mike McGrath
> <mmcgrath(a)redhat.com&gt; wrote:
> > On Tue, 24 Aug 2010, Till Maas wrote:
> >> If they like a faster bootup, then yes, they will. And I as a
> >> workstation user like it.
> >>
> >
> > [citation needed]
> >
> > I asked for this and was told by developers they were reluctant to post
> > the data.
>  
>  I also wonder who these people are who boot their computers so often
>  that boot-up speed makes any difference over a whole day. 
I reboot my computer ~5-10 times a day due to this bug:
https://bugzilla.redhat.com/show_bug.cgi?id=625187

Nathaniel

[Jon Masters]

On Wed, 2010-08-25 at 12:20 +0200, Christof Damian wrote:
>  On Tue, Aug 24, 2010 at 23:06, Mike McGrath
> <mmcgrath(a)redhat.com&gt; wrote:
>  > On Tue, 24 Aug 2010, Till Maas wrote:
>  >> If they like a faster bootup, then yes, they will. And I as a
>  >> workstation user like it.
>  >>
>  >
>  > [citation needed]
>  >
>  > I asked for this and was told by developers they were reluctant to post
>  > the data.
>  
>  I also wonder who these people are who boot their computers so often
>  that boot-up speed makes any difference over a whole day. 
Must be a limited set of developers. Let's face it, the average user
doesn't reboot all that often if they're doing suspend/resume, etc.
Those doing kernel development/plumbing hopefully have a separate test
box that isn't running a lot anyway (so won't take long to boot).

My desktop uptime:

[jcm@constitution ~]$ uptime
 09:28:22 up 24 days, 16:32,  9 users,  load average: 1.17, 0.50, 0.37

So clearly I'm absolutely, desperately concerned with 30 seconds or 5
minutes of boot time and it would be devastating if it took longer...

Jon.

[Matthew Garrett]

On Wed, Aug 25, 2010 at 09:31:30AM -0400, Jon Masters wrote:

>  [jcm@constitution ~]$ uptime
>   09:28:22 up 24 days, 16:32,  9 users,  load average: 1.17, 0.50, 0.37 
So you're running an insecure kernel? Our security churn is bad enough 
right now that rebooting is something people do fairly regularly.

-- 
Matthew Garrett | mjg59(a)srcf.ucam.org

[Jon Masters]

On Wed, 2010-08-25 at 18:37 +0100, Matthew Garrett wrote:
>  On Wed, Aug 25, 2010 at 09:31:30AM -0400, Jon Masters wrote:
>  
>  > [jcm@constitution ~]$ uptime
>  >  09:28:22 up 24 days, 16:32,  9 users,  load average: 1.17, 0.50, 0.37
>  
>  So you're running an insecure kernel? Our security churn is bad enough 
>  right now that rebooting is something people do fairly regularly. 
Yup, this desktop is running a kernel that still needs upgrading. I've
scheduled some time later this week to build a new one and test it
before I reboot the machine, but I won't be doing a yum update. When
I've got anecdotal evidence that updates have weeklong soak test time,
the barrier to entry is much higher, or similar, then I will do that.

Jon.

[Richard W.M. Jones]

On Wed, Aug 25, 2010 at 12:20:52PM +0200, Christof Damian wrote:
>  On Tue, Aug 24, 2010 at 23:06, Mike McGrath
> <mmcgrath(a)redhat.com&gt; wrote:
>  > On Tue, 24 Aug 2010, Till Maas wrote:
>  >> If they like a faster bootup, then yes, they will. And I as a
>  >> workstation user like it.
>  >>
>  >
>  > [citation needed]
>  >
>  > I asked for this and was told by developers they were reluctant to post
>  > the data.
>  
>  I also wonder who these people are who boot their computers so often
>  that boot-up speed makes any difference over a whole day. 
I reboot virtual machines all the time, certainly 100 times a day
would not be unusual on a work day.

Not an argument for or against systemd BTW, just an observation that
how you use your computer is not how others use their computers.  You
need to measure these things scientifically, not base it on your own
personal experience.

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
libguestfs lets you edit virtual machines.  Supports shell scripting,
bindings from many languages.  http://et.redhat.com/~rjones/libguestfs/
See what it can do: http://et.redhat.com/~rjones/libguestfs/recipes.html

[Jon Masters]

On Wed, 2010-08-25 at 17:06 +0100, Richard W.M. Jones wrote:

>  I reboot virtual machines all the time, certainly 100 times a day
>  would not be unusual on a work day.
>  
>  Not an argument for or against systemd BTW, just an observation that
>  how you use your computer is not how others use their computers.  You
>  need to measure these things scientifically, not base it on your own
>  personal experience. 
Both true. I also have dozens of virtual machines that are frequently
rebooted, but generally speaking I hit the virtual button and don't care
whether it takes 30, or 38 seconds to get there. It's like when watching
a server and often laughing at how long it'll take the BIOS and bootable
devices to do their thing compared with the eventual, actual boot.

Anyway, I've always read the boot time thing as just a bit of fun had by
those who probably notice it the most. I've yet to be convinced it
really matters to everyone else any more, but you're right about stats
being the answer. I'd love to see a Fedora User Survey about what they
actually care about, and I asked for one to be done some time ago.

Jon.

[Richard W.M. Jones]

On Wed, Aug 25, 2010 at 01:38:00PM -0400, Jon Masters wrote:
>  Anyway, I've always read the boot time thing as just a bit of fun
> had by
>  those who probably notice it the most. I've yet to be convinced it
>  really matters to everyone else any more, 
libguestfs cares (and you'll care if you're waiting for a libguestfs
command to run).  We don't run systemd or very much of userspace, but
we care how long it takes to boot KVM, the kernel and udevd.

I'm happy to say with the latest set of changes (not yet pushed out to
Fedora), this is down to under 5 seconds.

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
New in Fedora 11: Fedora Windows cross-compiler. Compile Windows
programs, test, and build Windows installers. Over 70 libraries supprt'd
http://fedoraproject.org/wiki/MinGW http://www.annexia.org/fedora_mingw

[drago01]

On Tue, Aug 24, 2010 at 11:06 PM, Mike McGrath <mmcgrath(a)redhat.com&gt; wrote:
>  On Tue, 24 Aug 2010, Till Maas wrote:
>
> > On Tue, Aug 24, 2010 at 10:46:41PM +0200, Farkas Levente wrote:
> >
> > > for workstation most users already use ubuntu. why? because it's more
> > > user friendly.
> >
> > There is nothing wrong with using Ubuntu, if it servers their needs.
> >
> > > do you think workstation users will like this kind of changes?
> >
> > If they like a faster bootup, then yes, they will. And I as a
> > workstation user like it.
> >
>
>  [citation needed]
>
>  I asked for this and was told by developers they were reluctant to post
>  the data. 
http://lwn.net/Articles/401441/

[dr johnson]

That would be all grand and spiffy if it were actually faster bootup.

Bootchart here reports that systemd is 8 seconds *SLOWER* than upstart.  No
idea why, but systemd just hangs for 8 seconds doing "nothing" that I can
see.  No logs anywhere that are meaningful.  Default clean install from
Alpha iso or Nightly livecd -- both do the same.

It is not booting faster or even the same speed.

I would like to see your bootchart for both if you report that systemd is
actually faster at all.

 -dj

ps: I should point out that it's 22 seconds to boot with upstart.  +8
seconds is a significant percentage increase.



-- 
   Scire ubi aliquid invenire possis, ea demum maxima pars eruditionis est
    --Anonymous


-----Original Message-----


On Tue, Aug 24, 2010 at 4:02 PM, Till Maas <opensource(a)till.name&gt; wrote:

>  On Tue, Aug 24, 2010 at 10:46:41PM +0200, Farkas Levente wrote:
>
>  > for workstation most users already use ubuntu. why? because it's more
>  > user friendly.
>
>  There is nothing wrong with using Ubuntu, if it servers their needs.
>
>  > do you think workstation users will like this kind of changes?
>
>  If they like a faster bootup, then yes, they will. And I as a
>  workstation user like it.
>
>  > do you think it'll be good to diverge from all other linux distros?
>
>  As long as we diverge in the direction of progress, then yes, it is. It
>  is a core principle of Fedora. Also this diversion is the only reason
>  why there are several distributions. The change might even drive more
>  users to Fedora.
>
>  > do you think workstation users will see any kind of uplift because of
>  > systemd?
>
>  Yes, they get a faster bootup afaik and happier package maintainers. ;-)
>
>  Regards
>  Till
>
>  --
>  devel mailing list
>  devel(a)lists.fedoraproject.org
>  https://admin.fedoraproject.org/mailman/listinfo/devel
>

[Bill Nottingham]

dr johnson (dj(a)www.uk.linux.org) said: 
>  That would be all grand and spiffy if it were actually faster
> bootup.
>  
>  Bootchart here reports that systemd is 8 seconds *SLOWER* than upstart.  No
>  idea why, but systemd just hangs for 8 seconds doing "nothing" that I can
>  see.  No logs anywhere that are meaningful.  Default clean install from
>  Alpha iso or Nightly livecd -- both do the same. 
Can you file this, please? (Apologies if I missed it.)

Bill

[dr johnson]

Certainly will after there is a new version pushed out.  Currently I'm
blocked by three other "fixed upstream" bugs that have been closed with no
update to f14 repos.

Maybe the new revision will solve the bootime issue too.

I am very curious if anyone can demonstrate that systemd is actually faster
than upstart at all. Even if only by one second.

 -dj



-- 
   Scire ubi aliquid invenire possis, ea demum maxima pars eruditionis est
    --Anonymous


-----Original Message-----


On Wed, Aug 25, 2010 at 2:52 PM, Bill Nottingham <notting(a)redhat.com&gt; wrote:

>  dr johnson (dj(a)www.uk.linux.org) said:
>  > That would be all grand and spiffy if it were actually faster bootup.
>  >
>  > Bootchart here reports that systemd is 8 seconds *SLOWER* than upstart.
>   No
>  > idea why, but systemd just hangs for 8 seconds doing "nothing" that I can
>  > see.  No logs anywhere that are meaningful.  Default clean install from
>  > Alpha iso or Nightly livecd -- both do the same.
>
>  Can you file this, please? (Apologies if I missed it.)
>
>  Bill
>  --
>  devel mailing list
>  devel(a)lists.fedoraproject.org
>  https://admin.fedoraproject.org/mailman/listinfo/devel
>

[Michael Schwendt]

On Tue, 24 Aug 2010 22:46:41 +0200, Farkas wrote:

>  and we all know what happened with pulseaudio. everybody turn it off
> and
>  remove it from the system in at least 3 fedora release, just because
>  some of their apps are not working. 
Which is a pain, admittedly, if those users are the developers of an audio
application even and don't make a good experience with Pulse Audio.
[The Audacious devs at some point suddenly had dropped the PA plugin in
favour of their completely rewritten ALSA output plugin, and only patches
for making it work with changes in PA could convince them to revive it.]

In Fedora currently something worse happens. And that is, Pulse Audio
development in Fedora 14 is non-existant, unfortunately,

  $ rpm -q pulseaudio
  pulseaudio-0.9.21-6.fc13.x86_64

because of systemd receiving the focus for several months, apparently.