FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor

Go Back   Linux Archive > Debian > Debian Development

 
 
LinkBack Thread Tools
 
Old 05-08-2010, 06:29 PM
Petter Reinholdtsen
 
Default Parallellizing the boot in Debian Squeeze - ready for wider testing

[Kai Wasserbäch]
> as one of the testers just a short reply: on several desktops and
> some basic servers insserv in conjunction with
> "CONCURRENCY=makefile" works well. I didn't have an unbootable
> system so far.

Thank you. Note that I do not expect an unbootable system. The worst
I expect are some services failing to start at boot, for example
because they are unable to connect to some other service or because a
file/socket/something isn't available at the right time during boot.
To fail completely to boot, the common packages would have to have
serious problems, and I am very confident that this is not the case.

> As far as speed goes: I didn't measure the speed but it feels
> faster.

Good to hear.

If the consensus is that we should switch the default to
CONCURRENCY=makefile, I guess it might still be time to do it.

After all, any bugs detected would be easy to work around, by editing
the dependency information in the files in /etc/init.d/. Any sysadmin
finding a problem could fix it locally while waiting for the fix to
show up.

Happy hacking,
--
Petter Reinholdtsen


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 2flbpcqckmw.fsf@login2.uio.no">http://lists.debian.org/2flbpcqckmw.fsf@login2.uio.no
 
Old 05-09-2010, 06:06 AM
Tollef Fog Heen
 
Default Parallellizing the boot in Debian Squeeze - ready for wider testing

]] (Marco d'Itri)

| Removing the Essential flag from sysvinit would allow interested admins
| to install upstart on their systems if they want to benefit from its
| features. I am not sure how much useful it would be to also switch to
| upstart by default in this scenario, I welcome other opinions.

I just filed an ITP on systemd and am planning on making it installable
alongside with sysvinit, switchable with init=/sbin/systemd when
booting. Eventually, I guess either using alternatives for /sbin/init
or dpkg-divert-ing it away from sysvinit might be the way to go. Either
way, whether sysvinit is Essential or not is a bit irrelevant to whether
people can experiment with other init systems.

--
Tollef Fog Heen
UNIX is user friendly, it's just picky about who its friends are


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 87k4rdvcbv.fsf@qurzaw.linpro.no">http://lists.debian.org/87k4rdvcbv.fsf@qurzaw.linpro.no
 
Old 05-09-2010, 08:01 AM
Stefano Zacchiroli
 
Default Parallellizing the boot in Debian Squeeze - ready for wider testing

On Sat, May 08, 2010 at 07:07:44PM +0200, Petter Reinholdtsen wrote:
> Perhaps you are right. Perhaps we should do a poll to collect
> information on how testers experience their boot with
> CONCURRENCY=makefile, to make it easier to switch with some confidence
> that it would work for most users.

Now you're talking!

> Well, it has not really been discussed with the release team, and the
> decision depend a lot of when Squeeze freezes, so it is hard to know
> what to decide.
>
> Perhaps we should switch the default in unstable to
> CONCURRENCY=makefile for a while, and if it causes a lot of problems
> we can switch it back to sequential boot. At the moment I believe we
> need to increase the amount of testing a lot to get the remaining bugs
> located and fixed, and that is hard to do without actually doing the
> switch.

If you are ready to monitor the issue closely, I don't see any problem
in switching the default now in unstable, see how it goes, and then
decide later on if revert back to the current default in Squeeze
time. Ideally, you should probably communicate a on the matter when/if
it arrives in testing to raise the awareness for testing users (and
probably we should work on the doc, as reported by Vincent). AFAICT the
change of the default change is relatively self-contained and will only
help in showing other problems which has been difficult to spot thus
far. [ I'm Cc-ing the release team, in case they see extra problems that
I don't in doing that change in the interim. ]


Cheers.

--
Stefano Zacchiroli -o- PhD in Computer Science PostDoc @ Univ. Paris 7
zack@{upsilon.cc,pps.jussieu.fr,debian.org} -<>- http://upsilon.cc/zack/
Dietro un grande uomo c'č ..| . |. Et ne m'en veux pas si je te tutoie
sempre uno zaino ...........| ..: |.... Je dis tu ā tous ceux que j'aime
 
Old 05-09-2010, 08:06 AM
Stefano Zacchiroli
 
Default Parallellizing the boot in Debian Squeeze - ready for wider testing

On Sun, May 09, 2010 at 08:06:12AM +0200, Tollef Fog Heen wrote:
> I just filed an ITP on systemd and am planning on

Amazing!, thanks for this.

> making it installable alongside with sysvinit, switchable with
> init=/sbin/systemd when booting. Eventually, I guess either using
> alternatives for /sbin/init or dpkg-divert-ing it away from sysvinit
> might be the way to go. Either way, whether sysvinit is Essential or
> not is a bit irrelevant to whether people can experiment with other
> init systems.

I've just read a few days ago the design document of systemd; AFAIU it
requires anyhow patching various daemons, no matter how trivial the
patches are.

Do you have any experience or feeling on how usable it would be today in
a Debian unstable box? Can we imagine having it as an alternative init
which a Squeeze user might want to try, obtaining something more than a
machine that simply doesn't boot half of the machine services?

Many thanks for your work on this,
Cheers.

--
Stefano Zacchiroli -o- PhD in Computer Science PostDoc @ Univ. Paris 7
zack@{upsilon.cc,pps.jussieu.fr,debian.org} -<>- http://upsilon.cc/zack/
Dietro un grande uomo c'č ..| . |. Et ne m'en veux pas si je te tutoie
sempre uno zaino ...........| ..: |.... Je dis tu ā tous ceux que j'aime
 
Old 05-09-2010, 10:26 AM
Cesare Leonardi
 
Default Parallellizing the boot in Debian Squeeze - ready for wider testing

On 08/05/2010 19:07, Petter Reinholdtsen wrote:

Perhaps we should do a poll to collect
information on how testers experience their boot with
CONCURRENCY=makefile, to make it easier to switch with some confidence
that it would work for most users.


If that helps, reading this thread i've set the previous variable in my
notebook (Sid with Gnome environment). I can see no problem but the
speed improvement is really small.


Here what i've measured, from the Grub start to the Gdm prompt, in
either case starting from a completely power off machine:

Without concurrency: 33 sec.
With concurrency (try 1): 29 sec.
With concurrency (try 2): 31 sec.

Bye.

Cesare.


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 4BE68DDC.2000609@gmail.com">http://lists.debian.org/4BE68DDC.2000609@gmail.com
 
Old 05-09-2010, 11:24 AM
Marc Haber
 
Default Parallellizing the boot in Debian Squeeze - ready for wider testing

On Thu, 06 May 2010 21:11:56 +0200, Petter Reinholdtsen
<pere@hungry.com> wrote:
>These days, the init.d script dependencies in Squeeze are quite
>complete, so complete that it is actually possible to run all the
>init.d scripts in parallell based on these dependencies. If you want
>to test your Squeeze system, make sure dependency based boot
>sequencing is enabled, and add this line to /etc/default/rcS:
>
> CONCURRENCY=makefile

Where is this documented?

Greetings
Marc
--
-------------------------------------- !! No courtesy copies, please !! -----
Marc Haber | " Questions are the | Mailadresse im Header
Mannheim, Germany | Beginning of Wisdom " | http://www.zugschlus.de/
Nordisch by Nature | Lt. Worf, TNG "Rightful Heir" | Fon: *49 621 72739834


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: E1OB4cr-0005Xg-KL@swivel.zugschlus.de">http://lists.debian.org/E1OB4cr-0005Xg-KL@swivel.zugschlus.de
 
Old 05-09-2010, 11:26 AM
Marc Haber
 
Default Parallellizing the boot in Debian Squeeze - ready for wider testing

On Sat, 8 May 2010 11:51:22 +0200, Stefano Zacchiroli
<zack@debian.org> wrote:
>On Sat, May 08, 2010 at 11:37:10AM +0200, Marc Haber wrote:
>> So it is the classical desktop vs. server situation. For my Debian
>> servers, that get booted at most once a month, I don't give a damn
>> about a faster boot.
>>
>> I _do_ care, however, about not having migrations in the boot process
>> which has the potential of breaking a system reboot, possibly making
>> it necessary to obtain a means to access the console in case of
>> breakage.
>
>Sure, but I don't think that the server argument should be use to imply
>that no changes, that benefit some of our users, are allowed at all.

No, but it is valid to avoid a migration of this magnitude weeks
before the freeze, with the next migration to the next new mechanism
already on the horizon.

Greetings
Marc
--
-------------------------------------- !! No courtesy copies, please !! -----
Marc Haber | " Questions are the | Mailadresse im Header
Mannheim, Germany | Beginning of Wisdom " | http://www.zugschlus.de/
Nordisch by Nature | Lt. Worf, TNG "Rightful Heir" | Fon: *49 621 72739834


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: E1OB4eA-0005fh-RU@swivel.zugschlus.de">http://lists.debian.org/E1OB4eA-0005fh-RU@swivel.zugschlus.de
 
Old 05-09-2010, 11:27 AM
Marc Haber
 
Default Parallellizing the boot in Debian Squeeze - ready for wider testing

On Sat, 8 May 2010 11:47:40 +0200, Julien Cristau
<jcristau@debian.org> wrote:
>As far as I'm concerned, "faster boot" is irrelevant. Using an init
>daemon that actually does its job of supervising services, and lets us
>get rid of most of the stupidity and boilerplate of init scripts, otoh,
>is overdue.

What is so bad about init scripts? Where am I supposed to put my init
script magic[1] in an upstart scenario?

Greetings
Marc

[1] a lot of my packages generate the configuration for the daemon
immediately before it is started
--
-------------------------------------- !! No courtesy copies, please !! -----
Marc Haber | " Questions are the | Mailadresse im Header
Mannheim, Germany | Beginning of Wisdom " | http://www.zugschlus.de/
Nordisch by Nature | Lt. Worf, TNG "Rightful Heir" | Fon: *49 621 72739834


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: E1OB4f6-0005gL-0J@swivel.zugschlus.de">http://lists.debian.org/E1OB4f6-0005gL-0J@swivel.zugschlus.de
 
Old 05-09-2010, 11:43 AM
Petter Reinholdtsen
 
Default Parallellizing the boot in Debian Squeeze - ready for wider testing

[Stefano Zacchiroli]
> If you are ready to monitor the issue closely, I don't see any problem
> in switching the default now in unstable, see how it goes, and then
> decide later on if revert back to the current default in Squeeze
> time. Ideally, you should probably communicate a on the matter when/if
> it arrives in testing to raise the awareness for testing users (and
> probably we should work on the doc, as reported by Vincent). AFAICT the
> change of the default change is relatively self-contained and will only
> help in showing other problems which has been difficult to spot thus
> far. [ I'm Cc-ing the release team, in case they see extra problems that
> I don't in doing that change in the interim. ]

I will continue to monitor this closesly, but do not have much time to
spend on it and thus need others to help out with providing patches
and fixing packages when problems are detected. And there are still
problems around. Just discovered that pdns-server do not start after
slapd when concurrent booting is enabled, and thus cases pdns to fail
when it is set up to use LDAP as its backend. I expect other similar
edge cases to be found as well, and my fear is that there will be a
lot of them that will not be detected before Squeeze is released,
causing the perceived quality of Debian to be reduced when Squeeze is
released.

Anyway, switching the default is a simple matter, editing a line in a
file in the sysv-rc package, and it would be quick both to enable and
disable if it proves to be a bad idea at the moment.

If the release team are OK with the switch, I assume the consensus in
this matter is that we should try to switch and see how it goes.

I am currently busy elsewhere, so it will at the earliest happen next
weekend.

Happy hacking,
--
Petter Reinholdtsen


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 2fl7hndi9lg.fsf@login2.uio.no">http://lists.debian.org/2fl7hndi9lg.fsf@login2.uio.no
 
Old 05-09-2010, 12:58 PM
Stefano Zacchiroli
 
Default Parallellizing the boot in Debian Squeeze - ready for wider testing

On Sun, May 09, 2010 at 01:24:49PM +0200, Marc Haber wrote:
> > CONCURRENCY=makefile
> Where is this documented?

According to <4BE43663.6000600@free.fr> and #576788, it is not.
But I'm sure Petter welcome patches on this.

Cheers.

--
Stefano Zacchiroli -o- PhD in Computer Science PostDoc @ Univ. Paris 7
zack@{upsilon.cc,pps.jussieu.fr,debian.org} -<>- http://upsilon.cc/zack/
Dietro un grande uomo c'č ..| . |. Et ne m'en veux pas si je te tutoie
sempre uno zaino ...........| ..: |.... Je dis tu ā tous ceux que j'aime
 

Thread Tools




All times are GMT. The time now is 05:51 PM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright Š2007 - 2008, www.linux-archive.org