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 07-01-2008, 11:29 AM
Josselin Mouette
 
Default Not stopping daemons, where are we?

Le mardi 01 juillet 2008 * 11:55 +0100, James Westby a écrit :
> Given that in runlevels 0 and 6 there is an init script that terminates
> all running processes it is a waste to run an init script for each
> process before that. In Ubuntu we remove the scripts that do nothing
> more than terminate a process, and rely on the final script to clean up
> the running processes.

I don’t think there has been any decision on this topic. I don’t have
anything against such a move, but I think we need a clear consensus on
whether this is a good idea.

--
.'`.
: :' : We are debian.org. Lower your prices, surrender your code.
`. `' We will add your hardware and software distinctiveness to
`- our own. Resistance is futile.
 
Old 07-01-2008, 11:44 AM
James Westby
 
Default Not stopping daemons, where are we?

On Tue, 2008-07-01 at 13:29 +0200, Josselin Mouette wrote:
> Le mardi 01 juillet 2008 * 11:55 +0100, James Westby a écrit :
> > Given that in runlevels 0 and 6 there is an init script that terminates
> > all running processes it is a waste to run an init script for each
> > process before that. In Ubuntu we remove the scripts that do nothing
> > more than terminate a process, and rely on the final script to clean up
> > the running processes.
>
> I don’t think there has been any decision on this topic. I don’t have
> anything against such a move, but I think we need a clear consensus on
> whether this is a good idea.
>

Originally Ubuntu took an approach "multiuser" that was not accepted in
to Debian. This approach has now been deprecated in favour of the one
that the patch I sent uses.

As I understand it, there is nothing magic about the approach taken, it
just doesn't install the symlinks for rc0.d and rc6.d, and expects that
the process will be cleaned up. It also reflects this in the LSB
headers, so systems which use that information should also do the same
thing.

https://wiki.ubuntu.com/Teardown
https://lists.ubuntu.com/archives/ubuntu-devel-announce/2008-June/000430.html
http://lists.debian.org/debian-devel/2008/01/msg00007.html

>From reviewing the last thread again I realise that my patch still stops
at 20 in runlevel one. Depending on the outcome of this discussion we
may want to discuss that for this specific case in the bug report.

Thanks,

James


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 07-01-2008, 12:42 PM
Bernd Eckenfels
 
Default Not stopping daemons, where are we?

In article <1214912655.6931.45.camel@flash> you wrote:
> As I understand it, there is nothing magic about the approach taken, it
> just doesn't install the symlinks for rc0.d and rc6.d, and expects that
> the process will be cleaned up. It also reflects this in the LSB
> headers, so systems which use that information should also do the same
> thing.

This is all fine, but what about daemons which have dependencies on remote
filesystems without declaration? Apache comes to mind, which often may run
on a Network filesystem. If killing all processes at the shutdown does not
observe the priority/ordering of the daemons this might kill the filesystem
thread before apache can write out pending data.

Gruss
Bernd


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 07-01-2008, 01:05 PM
James Westby
 
Default Not stopping daemons, where are we?

On Tue, 2008-07-01 at 14:42 +0200, Bernd Eckenfels wrote:
> In article <1214912655.6931.45.camel@flash> you wrote:
> > As I understand it, there is nothing magic about the approach taken, it
> > just doesn't install the symlinks for rc0.d and rc6.d, and expects that
> > the process will be cleaned up. It also reflects this in the LSB
> > headers, so systems which use that information should also do the same
> > thing.
>
> This is all fine, but what about daemons which have dependencies on remote
> filesystems without declaration? Apache comes to mind, which often may run
> on a Network filesystem. If killing all processes at the shutdown does not
> observe the priority/ordering of the daemons this might kill the filesystem
> thread before apache can write out pending data.

There is no such proposal to do this for all daemons. That's why the
defaults have not changed and the individual packages must do it.

Quoting https://wiki.ubuntu.com/Teardown which I referenced in my
previous mail:

Apache: Performs a controlled shut down of the running Apache web
server. While a web server is normally not likely to have unflushed
writes, modules such as mod_perl, mod_python and PHP might; so it's
important that we do allow a controlled shutdown.


Thanks,

James


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 07-01-2008, 01:38 PM
Bernd Eckenfels
 
Default Not stopping daemons, where are we?

In article <1214917547.6931.48.camel@flash> you wrote:
> There is no such proposal to do this for all daemons. That's why the
> defaults have not changed and the individual packages must do it.

Yes, but do you think individual packages can decide how the environment
they are running in might look like? I mean the pending-write case is the
most obvious. But what about resolver caches, VPNs and the like?

Gruss
Bernd


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 07-01-2008, 01:38 PM
Frans Pop
 
Default Not stopping daemons, where are we?

Le mardi 01 juillet 2008 11:55 +0100, James Westby a crit :
> Given that in runlevels 0 and 6 there is an init script that terminates
> all running processes it is a waste to run an init script for each
> process before that. In Ubuntu we remove the scripts that do nothing
> more than terminate a process, and rely on the final script to clean up
> the running processes.

I happened to see a similar bug filed against backuppc.

How many of these bugs have been filed?
Are you aware of the Debian policy regarding mass bug filing [1]?

IMO this is a subject that definitely should have been discussed on
d-devel _before_ the bugs were filed.

Cheers,
FJP

[1]http://www.debian.org/doc/developers-reference/ch-beyond-pkging.en.html#s-submit-many-bugs
 
Old 07-01-2008, 01:50 PM
James Westby
 
Default Not stopping daemons, where are we?

On Tue, 2008-07-01 at 15:38 +0200, Frans Pop wrote:
> I happened to see a similar bug filed against backuppc.
>
> How many of these bugs have been filed?
> Are you aware of the Debian policy regarding mass bug filing [1]?
>

I have filed two (from memory). I am aware of that policy.

> IMO this is a subject that definitely should have been discussed on
> d-devel _before_ the bugs were filed.
>

It was discussed. What sort of resolution would you like before these
bugs were filed?

If you mean that discussion of a proposed MBF should have happened then
I'm not sure that what I have done warrants that.

If someone was to collect all of the Ubuntu patches to do this and submit
them at once then it obviously would. However, Ubuntu has only just changed
to something that is acceptable to Debian and so most packages have not
been updated yet.


I'm not saying that there isn't a better way to go about this, but
in this case a suitable package can be changed unilaterally with no
effects elsewhere, so patches can be forwarded as and when they are
ready.

If someone is interested in seeing this happen then I would be happy
to co-ordinate such that we can update all the Ubuntu patches to be
acceptable to Debian, and then follow the procedures to propose
a large-scale change to Debian.

Thanks,

James


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 07-01-2008, 06:35 PM
Henrique de Moraes Holschuh
 
Default Not stopping daemons, where are we?

On Tue, 01 Jul 2008, Josselin Mouette wrote:
> Le mardi 01 juillet 2008 11:55 +0100, James Westby a crit :
> > Given that in runlevels 0 and 6 there is an init script that terminates
> > all running processes it is a waste to run an init script for each
> > process before that. In Ubuntu we remove the scripts that do nothing
> > more than terminate a process, and rely on the final script to clean up
> > the running processes.
>
> I don???t think there has been any decision on this topic. I don???t have
> anything against such a move, but I think we need a clear consensus on
> whether this is a good idea.

To put it buntly: it is a per-package choice. Some daemons benefit from
more time for cleanups, some initscripts do extra cleanup. While some
don't, and really have no reason to need anything else than the killall5
-TERM will do.

AFAIK, bugs were filled on packages that only simply stop daemons asking for
the maintainers to check if they could forego the stop hook on runlevels 0
and 6.

--
"One disk to rule them all, One disk to find them. One disk to bring
them all and in the darkness grind them. In the Land of Redmond
where the shadows lie." -- The Silicon Valley Tarot
Henrique Holschuh


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 07-01-2008, 10:35 PM
Steve Langasek
 
Default Not stopping daemons, where are we?

On Tue, Jul 01, 2008 at 03:38:20PM +0200, Bernd Eckenfels wrote:
> In article <1214917547.6931.48.camel@flash> you wrote:
> > There is no such proposal to do this for all daemons. That's why the
> > defaults have not changed and the individual packages must do it.

> Yes, but do you think individual packages can decide how the environment
> they are running in might look like?

... yes? Isn't that what a package maintainer is for?

> I mean the pending-write case is the most obvious. But what about resolver
> caches, VPNs and the like?

What kind of data loss do you expect to arise from shutting down a VPN
client without giving it time to save state?

--
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.
Ubuntu Developer http://www.debian.org/
slangasek@ubuntu.com vorlon@debian.org


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 07-02-2008, 12:27 AM
Bernd Eckenfels
 
Default Not stopping daemons, where are we?

In article <20080701223517.GB2670@dario.dodds.net> you wrote:
>> I mean the pending-write case is the most obvious. But what about resolver
>> caches, VPNs and the like?
>
> What kind of data loss do you expect to arise from shutting down a VPN
> client without giving it time to save state?

I dont expect any data loss - hopefully protocols are not that
optimistic/broken. But with unclean shutdown you can affect external parties
with unexected errors. Like resolver problems, user not found and similiar
problems.

Gruss
Bernd


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 

Thread Tools




All times are GMT. The time now is 11:05 AM.

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