FC5 and new init system

Hi all,

There has been a long discussion about the need of a new init system, in order 
to replace SysVinit. A wiki page and preliminary code has been written. Now 
the development seems to have stopped.
Are there any chances for the new init system (systemservices) to be included 
in FC5? If not: why?

regards
Giovanni

PS: what is the "status" of early login? seems a very similar story...

[dragoran]

Giovanni Biczó wrote:

> Hi all,
>
> There has been a long discussion about the need of a new init system, in order 
> to replace SysVinit. A wiki page and preliminary code has been written. Now 
> the development seems to have stopped.
> Are there any chances for the new init system (systemservices) to be included 
> in FC5? If not: why?
>
> regards
> Giovanni
>
> PS: what is the "status" of early login? seems a very similar story...
>
>   
>  
interessting question...
also will early-login be back in test1 ?
in fc4 this improved the boot time (until you see the desktop) by 30-40% 
on my box

[Harald Hoyer]

Giovanni Biczó wrote:
>  Hi all,
>  
>  There has been a long discussion about the need of a new init system, in order 
>  to replace SysVinit. A wiki page and preliminary code has been written. Now 
>  the development seems to have stopped.
>  Are there any chances for the new init system (systemservices) to be included 
>  in FC5? If not: why? 
Lack of response, noone seemed to be interested in systemservices.. so I stopped.

[Rahul Sundaram]

Hi

> >
> > There has been a long discussion about the need of a new init system, 
> > in order to replace SysVinit. A wiki page and preliminary code has 
> > been written. Now the development seems to have stopped.
> > Are there any chances for the new init system (systemservices) to be 
> > included in FC5? If not: why?
>
>
>  Lack of response, noone seemed to be interested in systemservices.. so 
>  I stopped.
>     
There is definitely a lot of interest from users as visible from 
repeated queries in this list. If the work you have done is packaged and 
pushed into rawhide or Fedora Extras,  we can get more feedback.

regards
Rahul

[Dimi Paun]

From: "Harald Hoyer" <harald(a)redhat.com&gt;
>  Lack of response, noone seemed to be interested in systemservices..
> so I 
stopped.

This is too bad -- I was pretty excited to see this development.
As for the lack of response, at least on my side, simply comes
from the avoidance of sending a "Yay!" to the list :)

-- 
Dimi Paun <dimi(a)lattica.com&gt;
Lattica, Inc.

[Kenneth Porter]

--On Monday, November 14, 2005 11:45 AM -0500 Dimi Paun <dimi(a)lattica.com&gt; 
wrote:

>  This is too bad -- I was pretty excited to see this development.
>  As for the lack of response, at least on my side, simply comes
>  from the avoidance of sending a "Yay!" to the list :) 
Same here. I figured the development went off-list and I didn't want to do 
an AOL "me-too". But if that's what it takes.... ;)

[dragoran]

Dimi Paun wrote:

> From: "Harald Hoyer" <harald(a)redhat.com&gt;
>   
>
> >Lack of response, noone seemed to be interested in systemservices.. so I
> >    
> >
> stopped.
>
> This is too bad -- I was pretty excited to see this development.
> As for the lack of response, at least on my side, simply comes
> from the avoidance of sending a "Yay!" to the list :)
>
>   
>  
same here....
can it be still be done before fc5 ?

[Chris Chabot]

Umm yes please, would love to see this in place and moving forward!


<quote who="dragoran">
>  Dimi Paun wrote:
>
> >From: "Harald Hoyer" <harald(a)redhat.com&gt;
> >
> >
> >>Lack of response, noone seemed to be interested in systemservices.. so I
> >>
> >>
> >stopped.
> >
> >This is too bad -- I was pretty excited to see this development.
> >As for the lack of response, at least on my side, simply comes
> >from the avoidance of sending a "Yay!" to the list :)
> >
> >
> >
>  same here....
>  can it be still be done before fc5 ?
>
>  --
>  fedora-devel-list mailing list
>  fedora-devel-list(a)redhat.com
>  https://www.redhat.com/mailman/listinfo/fedora-devel-list
>

[Nicolas Mailhot]

Same here. A lot of people were holding their breath for this, trying
not to scare you off - seems it didn't have the intended effect.

Will go AOL from now on ;)

-- 
Nicolas Mailhot

[Lamont R. Peterson]

On Monday 14 November 2005 02:24pm, Nicolas Mailhot wrote:
>  Same here. A lot of people were holding their breath for this,
> trying
>  not to scare you off - seems it didn't have the intended effect.
>
>  Will go AOL from now on ;) 
Or we could all just hit Harald's <harald(a)redhat.com&gt; email directly with our 
AOL "Me too"'s and spare the list. ;)

You'll get at least 7 votes from us at Guru Labs; as long as there is no XML 
in it.
-- 
Lamont R. Peterson <lamont(a)gurulabs.com&gt;
Senior Instructor
Guru Labs, L.C. [ http://www.GuruLabs.com/ ]

[Dax Kelson]

On Mon, 2005-11-14 at 14:56 -0700, Lamont R. Peterson wrote:
>  On Monday 14 November 2005 02:24pm, Nicolas Mailhot wrote:
>  > Same here. A lot of people were holding their breath for this, trying
>  > not to scare you off - seems it didn't have the intended effect.
>  >
>  > Will go AOL from now on ;)
>  
>  Or we could all just hit Harald's <harald(a)redhat.com&gt; email directly with our 
>  AOL "Me too"'s and spare the list. ;)
>  
>  You'll get at least 7 votes from us at Guru Labs; as long as there is no XML 
>  in it. 
Well, more accurately, as long as it doesn't suck. :)

The devil is in the design and implementation details ... yadda yadda.

Dax Kelson

[Tyler Larson]

Dax Kelson wrote:
>  On Mon, 2005-11-14 at 14:56 -0700, Lamont R. Peterson wrote:
> > On Monday 14 November 2005 02:24pm, Nicolas Mailhot wrote:
> >> Same here. A lot of people were holding their breath for this, trying
> >> not to scare you off - seems it didn't have the intended effect.
> >>
> >> Will go AOL from now on ;)
> > Or we could all just hit Harald's <harald(a)redhat.com&gt; email directly with
> our 
> > AOL "Me too"'s and spare the list. ;)
> >
> > You'll get at least 7 votes from us at Guru Labs; as long as there is no XML 
> > in it.
>  
>  Well, more accurately, as long as it doesn't suck. :)
>  
>  The devil is in the design and implementation details ... yadda yadda.
>  
>  Dax Kelson
>   
Which, of course, brings up the question of what the details actually are.
When we say, "replace SysVinit", is everyone actually talking about replacing
init, or do we think we're talking about replacing initscripts (which is an
entirely different beast altogether).

I would vote for leaving init alone, but doing a more dependency-based startup
procedure, kind of like *ahem* gentoo.

When its all said and done, however this is accomplished, I would really like
to see services start up in parallel. It would really help our boot-up time.
Even Windows does this... why not Fedora?

[Rahul Sundaram]

Hi

> Which, of course, brings up the question of what the details actually
> are.
> When we say, "replace SysVinit", is everyone actually talking about replacing
> init, or do we think we're talking about replacing initscripts (which is an
> entirely different beast altogether).
>
> I would vote for leaving init alone, but doing a more dependency-based startup
> procedure, kind of like *ahem* gentoo.
>  
The earlier discussion that happened in this list listed out the 
details. Check the archives. Also this page
http://fedoraproject.org/wiki/FCNewInit

regards
Rahul

[Ignacio Vazquez-Abrams]

On Mon, 2005-11-14 at 22:24 +0100, Nicolas Mailhot wrote:
>  Same here. A lot of people were holding their breath for this,
> trying
>  not to scare you off - seems it didn't have the intended effect.
>  
>  Will go AOL from now on ;) 
I'll throw my vote on as well, not because I have a (large) problem with
initscripts, but just for the coolness factor.

-- 
Ignacio Vazquez-Abrams <ivazquez(a)ivazquez.net&gt;
http://fedora.ivazquez.net/

gpg --keyserver hkp://subkeys.pgp.net --recv-key 38028b72

[Aurelien Bompard]

> > Will go AOL from now on ;)
>  
>  I'll throw my vote on as well, not because I have a (large) problem with
>  initscripts, but just for the coolness factor. 
Should we setup a wiki page where everybody could vote ? ;o)
Or a simple slashdot-like web poll system ? (that could actually be useful)

Anyway, you have my vote too.

Aurélien
-- 
http://aurelien.bompard.org  ~~~~  Jabber : abompard(a)jabber.fr
No, I coded it crappily on purpose, just so that I could say "There's
plenty of room for optimization."

[Rahul Sundaram]

Aurelien Bompard wrote:

> >>Will go AOL from now on ;)
> >>      
> >>
> >I'll throw my vote on as well, not because I have a (large) problem with
> >initscripts, but just for the coolness factor.
> >    
> >
>
> Should we setup a wiki page where everybody could vote ? ;o)
> Or a simple slashdot-like web poll system ? (that could actually be useful)
>  
Helping in coding, testing, documentation and feedback helps drive the 
changes much more than merely voting for such changes.


regards
Rahul

[Aurelien Bompard]

Rahul Sundaram wrote:
>  Helping in coding, testing, documentation and feedback helps drive
> the
>  changes much more than merely voting for such changes. 
Agreed, but this thread proves that letting people express their
support/enthusiasm for a project can be useful too.

Aurélien.
-- 
http://aurelien.bompard.org  ~~~~  Jabber : abompard(a)jabber.fr
A Black Hole is where God divided by zero.

[Paul A Houle]

Ignacio Vazquez-Abrams wrote:

> I'll throw my vote on as well, not because I have a (large) problem with
> initscripts, but just for the coolness factor.
>   
>  
    I'd love to see an init system that can do parallel boot and service 
monitoring.

    At some point,  I often end up installing djb's daemontools on 
machines that I administer.  Sometimes this is because I install other 
djb software,  sometimes because I need the service monitoring.

    On the other hand,  a bad init system could make my life miserable,  
so I'd move carefully in this department.  I've been hearing bad things 
about the system in Solaris 10,  for instance.  I'd hate to have to 
write new init scripts for all the software I run -- even though 
dependency awareness is necessary for parallel boot,  I'd want the 
ability to break the dependency rules when I want to break them.

[Luke Macken]

On Sat, Nov 12, 2005 at 11:34:57AM +0000, Giovanni Biczó wrote:
| Hi all,
| 
| There has been a long discussion about the need of a new init system, in order 
| to replace SysVinit. A wiki page and preliminary code has been written. Now 
| the development seems to have stopped.
| Are there any chances for the new init system (systemservices) to be included 
| in FC5? If not: why?

The discussions for Fedora's SysVinit replacement are currently off-list,
but not for much longer.  Any inclusion of a new init system in FC5 isn't
going to happen considering today is the test1 devel freeze, but Fedora
Extras will most likely serve to be a good testbed for any init replacements
that we plan on testing in the near future.

luke

[Jesse Keating]

On Mon, 2005-11-14 at 13:08 -0500, Luke Macken wrote:
>  The discussions for Fedora's SysVinit replacement are currently
> off-list,
>  but not for much longer.  Any inclusion of a new init system in FC5 isn't
>  going to happen considering today is the test1 devel freeze, but Fedora
>  Extras will most likely serve to be a good testbed for any init replacements
>  that we plan on testing in the near future. 
Unfortunately Extras may not be the best place.  One of the rules of
Extras is that it doesn't replace something in Core.

Would the new init system require the SysV init stuff be
disabled/removed?

-- 
Jesse Keating RHCE      (http://geek.j2solutions.net)
Fedora Legacy Team      (http://www.fedoralegacy.org)
GPG Public Key
(http://geek.j2solutions.net/jkeating.j2solutions.pub)

Was I helpful?  Let others know:
 http://svcs.affero.net/rm.php?r=jkeating

[seth vidal]

On Mon, 2005-11-14 at 11:39 -0800, Jesse Keating wrote:
>  On Mon, 2005-11-14 at 13:08 -0500, Luke Macken wrote:
>  > The discussions for Fedora's SysVinit replacement are currently off-list,
>  > but not for much longer.  Any inclusion of a new init system in FC5 isn't
>  > going to happen considering today is the test1 devel freeze, but Fedora
>  > Extras will most likely serve to be a good testbed for any init replacements
>  > that we plan on testing in the near future.
>  
>  Unfortunately Extras may not be the best place.  One of the rules of
>  Extras is that it doesn't replace something in Core.
>  
>  Would the new init system require the SysV init stuff be
>  disabled/removed? 
well things in extras can be replacements for core but they cannot:
 - explicitly conflict or file-conflict
 - explicitly obsolete

but syslog-ng is in extras and it can serve as a replacement for syslogd

-sv

[Luke Macken]

On Mon, Nov 14, 2005 at 11:39:07AM -0800, Jesse Keating wrote:
| On Mon, 2005-11-14 at 13:08 -0500, Luke Macken wrote:
| > The discussions for Fedora's SysVinit replacement are currently off-list,
| > but not for much longer.  Any inclusion of a new init system in FC5 isn't
| > going to happen considering today is the test1 devel freeze, but Fedora
| > Extras will most likely serve to be a good testbed for any init replacements
| > that we plan on testing in the near future.
| 
| Unfortunately Extras may not be the best place.  One of the rules of
| Extras is that it doesn't replace something in Core.
| 
| Would the new init system require the SysV init stuff be
| disabled/removed?

That all depends on what we plan to implement.  Harald's SystemManager
is an /etc/rc replacement, and works along side of our SysVinit (iirc);
but something like initng[0] is a full out replacement, but has no
file conflicts and should work with SysVinit installed.  I originally
thought FE would be ideal because it would give anyone a chance to test
things out (although, SystemManager seems to be more transparent and
easier to toss in to core right away), and we wouldn't have to have
it obsolete anything until it gets into core.

luke

[0]: http://initng.thinktux.net