Bug#595964: Please cherry-pick patch to mount cgroupfs at /sys/fs/cgroup

2010-09-14 Thread Jon Bernard
* Michael Biebl bi...@debian.org wrote:
 CCing Jon, as maintainer of libcgroup he might be interested in this
 discussion.
 
 Am 07.09.2010 23:12, schrieb Bastian Blank:
  On Tue, Sep 07, 2010 at 10:54:04PM +0200, Michael Biebl wrote:
  On 07.09.2010 22:41, Bastian Blank wrote:
  And what does this help? There are n cgroup filesystem hierarchies.
  Are you referring to cgroup fs like cpuset? They would be mounted below
  /sys/fs/cgroup.
  
  sysfs does not support mkdir. How will you mount _n_ different
  hierarchies?
 
 In the thread I quoted, G-K H mentions that /sys/fs/cgroup would be a tmpfs.

I agree with not creating another top-level directory. I configured
libcgroup to mount cgroupfs under /mnt/cgroup as an initial solution,
but mounting in /sys/fs seems much more consistent with current
conventions.

  As it is private to systemd anyway, the correct locations is somewhere
  below /var/lib/
  As systemd requires the cgroupfs during early boot, /var/lib is not an 
  option.
  
  There are other locations. /sys is system wide and not daemon specific.
  
 
 I remember discussions, where to mount the cgroupfs [1], some while ago,
 and /sys was among the recommendations.  More importantly though, it was
 said that we should see what the kernel folks do.

This discussion took place on the libcgroup mailing list as well [1],
posting here for reference. Many suggested somewhere under /sys, but I
don't believe a consensus was reached.

[1]: 
http://sourceforge.net/mailarchive/message.php?msg_id=20091104063005.GC3560%40balbir.in.ibm.com

Cheers

-- 
Jon



-- 
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/20100914135146.ga1...@starquasia



Bug#595964: Please cherry-pick patch to mount cgroupfs at /sys/fs/cgroup

2010-09-14 Thread Ben Hutchings
On Tue, Sep 14, 2010 at 09:51:49AM -0400, Jon Bernard wrote:
 * Michael Biebl bi...@debian.org wrote:
  CCing Jon, as maintainer of libcgroup he might be interested in this
  discussion.
  
  Am 07.09.2010 23:12, schrieb Bastian Blank:
   On Tue, Sep 07, 2010 at 10:54:04PM +0200, Michael Biebl wrote:
   On 07.09.2010 22:41, Bastian Blank wrote:
   And what does this help? There are n cgroup filesystem hierarchies.
   Are you referring to cgroup fs like cpuset? They would be mounted below
   /sys/fs/cgroup.
   
   sysfs does not support mkdir. How will you mount _n_ different
   hierarchies?
  
  In the thread I quoted, G-K H mentions that /sys/fs/cgroup would be a tmpfs.
 
 I agree with not creating another top-level directory. I configured
 libcgroup to mount cgroupfs under /mnt/cgroup as an initial solution,
 but mounting in /sys/fs seems much more consistent with current
 conventions.

Using /mnt/cgroup is definitely wrong.  FHS says that /mnt is reserved as a
temporary mountpoint for local use.

If there is a consensus upstream that /sys/fs/cgroup is the right place for
this filesystem, and this is needed in squeeze (rather than just for
systemd) then I think we would add that directory in the kernel package for
squeeze.  Otherwise, this is for wheezy (and current experimental) only.

Ben.

-- 
Ben Hutchings
We get into the habit of living before acquiring the habit of thinking.
  - Albert Camus



-- 
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/20100914145957.gf11...@decadent.org.uk



Bug#595964: Please cherry-pick patch to mount cgroupfs at /sys/fs/cgroup

2010-09-14 Thread Michael Biebl
On 14.09.2010 16:59, Ben Hutchings wrote:
 On Tue, Sep 14, 2010 at 09:51:49AM -0400, Jon Bernard wrote:


 I agree with not creating another top-level directory. I configured
 libcgroup to mount cgroupfs under /mnt/cgroup as an initial solution,
 but mounting in /sys/fs seems much more consistent with current
 conventions.
 
 Using /mnt/cgroup is definitely wrong.  FHS says that /mnt is reserved as a
 temporary mountpoint for local use.
 
 If there is a consensus upstream that /sys/fs/cgroup is the right place for
 this filesystem,

This patch has ben acked by
Paul Menage men...@google.com and
Dhaval Giani dhaval.gi...@gmail.com

Afaik Dhaval is the libcgroup upstream developer and Paul is the maintainer of
the kernel cgroup code. And also as mentioned before, it will be in 2.6.36. I
guess this can be considered as consensus upstream.

 and this is needed in squeeze (rather than just for
 systemd) then I think we would add that directory in the kernel package for
 squeeze.  Otherwise, this is for wheezy (and current experimental) only.

As we usually try to support running the old kernel from release (squeeze) in
release+1(wheezy), this patch would ensure that systemd would easily run with a
squeeze kernel (without any ugly workarounds). It would be future-proofing if
you will.

Cheers,
Michael

-- 
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?



signature.asc
Description: OpenPGP digital signature


Bug#595964: Please cherry-pick patch to mount cgroupfs at /sys/fs/cgroup

2010-09-14 Thread Ben Hutchings
On Tue, Sep 14, 2010 at 06:29:10PM +0200, Michael Biebl wrote:
 On 14.09.2010 16:59, Ben Hutchings wrote:
[...]
  and this is needed in squeeze (rather than just for
  systemd) then I think we would add that directory in the kernel package for
  squeeze.  Otherwise, this is for wheezy (and current experimental) only.
 
 As we usually try to support running the old kernel from release (squeeze) in
 release+1(wheezy), this patch would ensure that systemd would easily run with 
 a
 squeeze kernel (without any ugly workarounds). It would be future-proofing if
 you will.

Who are we?  You'd better tell Md about this new requirement.

Ben.

-- 
Ben Hutchings
We get into the habit of living before acquiring the habit of thinking.
  - Albert Camus



-- 
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/20100914165724.gg11...@decadent.org.uk



Bug#595964: Please cherry-pick patch to mount cgroupfs at /sys/fs/cgroup

2010-09-14 Thread Tollef Fog Heen
]] Ben Hutchings 

| Who are we?  You'd better tell Md about this new requirement.

I, as the systemd maintainer would be much happier if it was possible to
use wheezy's systemd with the lenny kernel, and with the (arguably
large) assumption that it's just the mount point for cgroupfs, that
doesn't sound like a huge extra burden to add to the kernel maintenance?

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



-- 
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/87hbhstdce@qurzaw.linpro.no



Bug#595964: Please cherry-pick patch to mount cgroupfs at /sys/fs/cgroup

2010-09-07 Thread Michael Biebl
Package: linux-2.6
Version: 2.6.32-21
Severity: wishlist
File: /boot/vmlinuz-2.6.32-5-686

Hi,

please consider cherry-picking the patch to mount cgroupfs at
/sys/fs/cgroup [1]. This patch will be in upstream 2.6.36 and has been
acked by the kernel cgroup and libcgroup maintainer.
Having it in the default Debian kernel would simplify running systemd, which
uses cgroups to monitor processes.
Without this patch, the cgroupfs for systemd will be mounted at /cgroup
and it would be preferable to not introduce another toplevel directory.

Thanks for considering,
Michael


[1]
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=676db4af043014e852f67ba0349dae0071bd11f3

-- System Information:
Debian Release: squeeze/sid
  APT prefers unstable
  APT policy: (500, 'unstable'), (1, 'experimental')
Architecture: i386 (i686)

Kernel: Linux 2.6.32-5-686 (SMP w/1 CPU core)
Locale: LANG=de_DE.utf8, LC_CTYPE=de_DE.utf8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash

Versions of packages linux-image-2.6.32-5-686 depends on:
ii  debconf [debconf-2.0] 1.5.35 Debian configuration management sy
ii  initramfs-tools [linux-initra 0.98.2 tools for generating an initramfs
ii  linux-base2.6.32-21  Linux image base package
ii  module-init-tools 3.12-1 tools for managing Linux kernel mo

Versions of packages linux-image-2.6.32-5-686 recommends:
ii  firmware-linux-free   2.6.32-21  Binary firmware for various driver
ii  libc6-i6862.11.2-5   Embedded GNU C Library: Shared lib

Versions of packages linux-image-2.6.32-5-686 suggests:
pn  grub | lilo   none (no description available)
pn  linux-doc-2.6.32  none (no description available)

Versions of packages linux-image-2.6.32-5-686 is related to:
pn  firmware-bnx2 none (no description available)
pn  firmware-bnx2xnone (no description available)
ii  firmware-ipw2x00  0.26   Binary firmware for Intel Pro Wire
pn  firmware-ivtv none (no description available)
pn  firmware-iwlwifi  none (no description available)
ii  firmware-linux0.26   Binary firmware for various driver
ii  firmware-linux-nonfree0.26   Binary firmware for various driver
pn  firmware-qlogic   none (no description available)
pn  firmware-ralink   none (no description available)
pn  xen-hypervisornone (no description available)

-- debconf information excluded



-- 
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Archive: 
http://lists.debian.org/20100907171915.2063.91589.report...@pluto.milchstrasse.xx



Bug#595964: Please cherry-pick patch to mount cgroupfs at /sys/fs/cgroup

2010-09-07 Thread Bastian Blank
On Tue, Sep 07, 2010 at 07:19:15PM +0200, Michael Biebl wrote:
 please consider cherry-picking the patch to mount cgroupfs at
 /sys/fs/cgroup [1]. This patch will be in upstream 2.6.36 and has been
 acked by the kernel cgroup and libcgroup maintainer.

And what does this help? There are n cgroup filesystem hierarchies.

 Without this patch, the cgroupfs for systemd will be mounted at /cgroup
 and it would be preferable to not introduce another toplevel directory.

As it is private to systemd anyway, the correct locations is somewhere
below /var/lib/

Bastian

-- 
There's another way to survive.  Mutual trust -- and help.
-- Kirk, Day of the Dove, stardate unknown



-- 
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/20100907204131.ga22...@wavehammer.waldi.eu.org



Bug#595964: Please cherry-pick patch to mount cgroupfs at /sys/fs/cgroup

2010-09-07 Thread Michael Biebl
On 07.09.2010 22:41, Bastian Blank wrote:
 On Tue, Sep 07, 2010 at 07:19:15PM +0200, Michael Biebl wrote:
 please consider cherry-picking the patch to mount cgroupfs at
 /sys/fs/cgroup [1]. This patch will be in upstream 2.6.36 and has been
 acked by the kernel cgroup and libcgroup maintainer.
 
 And what does this help? There are n cgroup filesystem hierarchies.

Are you referring to cgroup fs like cpuset? They would be mounted below
/sys/fs/cgroup.

One idea behind this patch is, to provide a sane default by the kernel, so the
userland does not mount it at random places, like /dev/cgroup, /cgroup or
/mnt/cgroup.

See also the discussion at [1].

 
 Without this patch, the cgroupfs for systemd will be mounted at /cgroup
 and it would be preferable to not introduce another toplevel directory.
 
 As it is private to systemd anyway, the correct locations is somewhere
 below /var/lib/

As systemd requires the cgroupfs during early boot, /var/lib is not an option.

Michael


[1] http://lkml.org/lkml/2010/7/22/384
-- 
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?



signature.asc
Description: OpenPGP digital signature


Bug#595964: Please cherry-pick patch to mount cgroupfs at /sys/fs/cgroup

2010-09-07 Thread Bastian Blank
On Tue, Sep 07, 2010 at 10:54:04PM +0200, Michael Biebl wrote:
 On 07.09.2010 22:41, Bastian Blank wrote:
  And what does this help? There are n cgroup filesystem hierarchies.
 Are you referring to cgroup fs like cpuset? They would be mounted below
 /sys/fs/cgroup.

sysfs does not support mkdir. How will you mount _n_ different
hierarchies?

  As it is private to systemd anyway, the correct locations is somewhere
  below /var/lib/
 As systemd requires the cgroupfs during early boot, /var/lib is not an option.

There are other locations. /sys is system wide and not daemon specific.

Bastian

-- 
Intuition, however illogical, is recognized as a command prerogative.
-- Kirk, Obsession, stardate 3620.7



-- 
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/20100907211225.ga23...@wavehammer.waldi.eu.org



Bug#595964: Please cherry-pick patch to mount cgroupfs at /sys/fs/cgroup

2010-09-07 Thread Michael Biebl
CCing Jon, as maintainer of libcgroup he might be interested in this
discussion.

Am 07.09.2010 23:12, schrieb Bastian Blank:
 On Tue, Sep 07, 2010 at 10:54:04PM +0200, Michael Biebl wrote:
 On 07.09.2010 22:41, Bastian Blank wrote:
 And what does this help? There are n cgroup filesystem hierarchies.
 Are you referring to cgroup fs like cpuset? They would be mounted below
 /sys/fs/cgroup.
 
 sysfs does not support mkdir. How will you mount _n_ different
 hierarchies?

In the thread I quoted, G-K H mentions that /sys/fs/cgroup would be a tmpfs.

 As it is private to systemd anyway, the correct locations is somewhere
 below /var/lib/
 As systemd requires the cgroupfs during early boot, /var/lib is not an 
 option.
 
 There are other locations. /sys is system wide and not daemon specific.
 

I remember discussions, where to mount the cgroupfs [1], some while ago,
and /sys was among the recommendations.  More importantly though, it was
said that we should see what the kernel folks do.

With the aforementioned patch, the kernel maintainers have more or less
answered the question and as already said, this patch will land in .36
and the Debian kernel for that matter.

Do you see any problems, why this patch should not be in the .32
(squeeze) kernel?

Michael

[1] http://lists.debian.org/debian-devel/2009/02/msg00037.html
-- 
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?



signature.asc
Description: OpenPGP digital signature