[gentoo-project] Council meeting: Tuesday 3 April 2012, 19:00 UTC

[gentoo-project] Council meeting: Tuesday 3 April 2012, 19:00 UTC

[Ulrich Mueller]

The next council meeting will be on Tuesday 3 April 2012 at 19:00 UTC
(note change of time) [1] in the #gentoo-council channel on Freenode.

Proposed agenda:

1. Introduction and roll call (5 minutes)

2. EAPI specification in ebuilds (30 minutes)

   See [2]. Discuss/vote which of the proposals [3] should be further
   pursued and worked out in detail:
     A: Parse bash assignment statement
     B: Comment in first line of header
     C: Bash function
     D: Filename extension (GLEP 55)
     E: External metadata file
     F: None, keep status quo

3. New udev and separate /usr partition (30 minutes)

   See [4]: "Decide on whether a separate /usr is still a supported
   configuration. If it is, newer udev can not be stabled and
   alternatives should be investigated. If it isn't, a lot of
   documentation will have to be updated. (And an alternative should
   likely still be provided.)"

4. herds.xml and mail aliases (10 minutes)

   See [5]. Discuss: Should devs be required to add themselves to
   herds.xml when adding to mail alias?

5. Open bugs with council involvement (10 minutes)

     - Bug 383467 "Council webpage lacks results for 2010 and 2011
       elections" [6]
     - Bug 408073 "Make the user choose a locale" [7]

6. Open floor

Regards,
Ulrich

[1] <http://www.gentoo.org/proj/en/council/utctolocal.html?time=1900>
[2] <http://archives.gentoo.org/gentoo-dev/msg_288fa260bf4c9242a33380fd4bfce0f8.xml>
[3] <http://wiki.gentoo.org/wiki/Alternate_EAPI_mechanisms>
[4] <http://archives.gentoo.org/gentoo-project/msg_c96d1b724cd736702820fa5ff1547557.xml>
[5] <http://archives.gentoo.org/gentoo-project/msg_7c56d79cadae396581807f1632471c29.xml>
[6] <https://bugs.gentoo.org/show_bug.cgi?id=383467>
[7] <https://bugs.gentoo.org/show_bug.cgi?id=408073>

Re: [gentoo-project] Council meeting: Tuesday 3 April 2012, 19:00 UTC

[Sven Vermeulen]

On Tue, Mar 27, 2012 at 09:50:19AM +0200, Ulrich Mueller wrote:
> 3. New udev and separate /usr partition (30 minutes)
> 
>    See [4]: "Decide on whether a separate /usr is still a supported
>    configuration. If it is, newer udev can not be stabled and
>    alternatives should be investigated. If it isn't, a lot of
>    documentation will have to be updated. (And an alternative should
>    likely still be provided.)"

Is it truely that much documentation? The handbook doesn't suggest separate
/usr anymore (all example partitions use a single / file system), and the
necessary changes to setup an initramfs are being prepared (cfr bugs related
to bug #407959).

There will be some updates in the quickinstall for raid/lvm, but that's
hardly a minor update to the document. 

We might want to provide an initramfs-guide to explain the concepts and the
various methods for generating one (as well as debugging), but that too
might not be the most difficult thing out there (there's still time for
that, no)?

Wkr,
	Sven Vermeulen

Re: [gentoo-project] Council meeting: Tuesday 3 April 2012, 19:00 UTC

[Pandu Poluan]

[-- Attachment #1: Type: text/plain, Size: 1544 bytes --]

On Mar 27, 2012 10:43 PM, "Sven Vermeulen" <swift@gentoo.org> wrote:
>
> On Tue, Mar 27, 2012 at 09:50:19AM +0200, Ulrich Mueller wrote:
> > 3. New udev and separate /usr partition (30 minutes)
> >
> >    See [4]: "Decide on whether a separate /usr is still a supported
> >    configuration. If it is, newer udev can not be stabled and
> >    alternatives should be investigated. If it isn't, a lot of
> >    documentation will have to be updated. (And an alternative should
> >    likely still be provided.)"
>
> Is it truely that much documentation? The handbook doesn't suggest
separate
> /usr anymore (all example partitions use a single / file system), and the
> necessary changes to setup an initramfs are being prepared (cfr bugs
related
> to bug #407959).
>
> There will be some updates in the quickinstall for raid/lvm, but that's
> hardly a minor update to the document.
>
> We might want to provide an initramfs-guide to explain the concepts and
the
> various methods for generating one (as well as debugging), but that too
> might not be the most difficult thing out there (there's still time for
> that, no)?
>
> Wkr,
>        Sven Vermeulen
>

Alternatively, suggest using mdev in place of udev. So, there will be 3
options:

1) /usr on the same partition as / -- can use mdev or udev

2) /usr on a separate partition; this leads to :

2a) If one wants to use udev, one must use initramfs

2b) If one does not want to use an initramfs, one must use mdev

This will give the user a CHOICE, and that's the essence of Gentoo.

Rgds,

[-- Attachment #2: Type: text/html, Size: 1943 bytes --]

Re: [gentoo-project] Council meeting: Tuesday 3 April 2012, 19:00 UTC

[Zac Medico]

On 03/27/2012 09:03 AM, Pandu Poluan wrote:
> Alternatively, suggest using mdev in place of udev. So, there will be 3
> options:
>
> 1) /usr on the same partition as / -- can use mdev or udev
>
> 2) /usr on a separate partition; this leads to :
>
> 2a) If one wants to use udev, one must use initramfs
>
> 2b) If one does not want to use an initramfs, one must use mdev

I'd leave option 2b undocumented. It's fragile and it will only work up 
until something else breaks, and then we'll have another big "OMG 
they're forcing us to use an initramfs" discussion and have to update 
the documentation again.
-- 
Thanks,
Zac

Re: [gentoo-project] Council meeting: Tuesday 3 April 2012, 19:00 UTC

[Joshua Kinard]

[-- Attachment #1: Type: text/plain, Size: 697 bytes --]

On 03/27/2012 11:42, Sven Vermeulen wrote:

> 
> Is it truely that much documentation? The handbook doesn't suggest separate
> /usr anymore (all example partitions use a single / file system), and the
> necessary changes to setup an initramfs are being prepared (cfr bugs related
> to bug #407959).


Almost, see Code Listing 1.1:
http://www.gentoo.org/doc/en/security/security-handbook.xml?part=1&chap=4

-- 
Joshua Kinard
Gentoo/MIPS
kumba@gentoo.org
4096R/D25D95E3 2011-03-28

"The past tempts us, the present confuses us, the future frightens us.  And
our lives slip away, moment by moment, lost in that vast, terrible in-between."

--Emperor Turhan, Centauri Republic


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 834 bytes --]

Re: [gentoo-project] Council meeting: Tuesday 3 April 2012, 19:00 UTC

[Joshua Kinard]

[-- Attachment #1: Type: text/plain, Size: 1112 bytes --]

On 03/27/2012 12:23, Zac Medico wrote:

>> 2b) If one does not want to use an initramfs, one must use mdev
> 
> I'd leave option 2b undocumented. It's fragile and it will only work up
> until something else breaks, and then we'll have another big "OMG they're
> forcing us to use an initramfs" discussion and have to update the
> documentation again.


Not undocumented, but with a link to this Wiki entry:
https://wiki.gentoo.org/wiki/Mdev

And a BIG FAT WARNING that it's risky, and only suitable for extremely
simple system builds.  It works fine for me on my main dev box and a VM.
I'll know how it works on MIPS when I get that machine properly updated.

Doesn't mean it'll work for everyone, but we're about choice, and we should
at least put a mention in, even if it's a choice that will be rarely taken.

-- 
Joshua Kinard
Gentoo/MIPS
kumba@gentoo.org
4096R/D25D95E3 2011-03-28

"The past tempts us, the present confuses us, the future frightens us.  And
our lives slip away, moment by moment, lost in that vast, terrible in-between."

--Emperor Turhan, Centauri Republic


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 834 bytes --]

Re: [gentoo-project] Council meeting: Tuesday 3 April 2012, 19:00 UTC

[vivo75]

Il 28/03/2012 01:18, Joshua Kinard ha scritto:
> On 03/27/2012 11:42, Sven Vermeulen wrote:
>
>> Is it truely that much documentation? The handbook doesn't suggest separate
>> /usr anymore (all example partitions use a single / file system), and the
>> necessary changes to setup an initramfs are being prepared (cfr bugs related
>> to bug #407959).
>
> Almost, see Code Listing 1.1:
> http://www.gentoo.org/doc/en/security/security-handbook.xml?part=1&chap=4
>
while there, may really be a good idea to replace reiserfs with whatever 
else

Re: [gentoo-project] Council meeting: Tuesday 3 April 2012, 19:00 UTC

[Pandu Poluan]

[-- Attachment #1: Type: text/plain, Size: 1389 bytes --]

On Mar 28, 2012 6:23 AM, "Joshua Kinard" <kumba@gentoo.org> wrote:
>
> On 03/27/2012 12:23, Zac Medico wrote:
>
> >> 2b) If one does not want to use an initramfs, one must use mdev
> >
> > I'd leave option 2b undocumented. It's fragile and it will only work up
> > until something else breaks, and then we'll have another big "OMG
they're
> > forcing us to use an initramfs" discussion and have to update the
> > documentation again.
>
>
> Not undocumented, but with a link to this Wiki entry:
> https://wiki.gentoo.org/wiki/Mdev
>
> And a BIG FAT WARNING that it's risky, and only suitable for extremely
> simple system builds.  It works fine for me on my main dev box and a VM.
> I'll know how it works on MIPS when I get that machine properly updated.
>
> Doesn't mean it'll work for everyone, but we're about choice, and we
should
> at least put a mention in, even if it's a choice that will be rarely
taken.
>

"Simple" is relative. I tend to say "non-exotic".

mdev is perfectly usable on my virtualized servers, whose devices are both
non-exotic (from the VM's point of view) and static.

Just recently, I polled the server mailing list, and while there's no clear
consensus, there are responders don't like option 2a because it involves an
initramfs.

But the essence is exactly what Joshua wrote in his last paragraph: it's
about choice. The user's (or sysadmin's) choice.

Rgds,

[-- Attachment #2: Type: text/html, Size: 1830 bytes --]

Re: [gentoo-project] Council meeting: Tuesday 3 April 2012, 19:00 UTC

[Joshua Kinard]

[-- Attachment #1: Type: text/plain, Size: 980 bytes --]

On 03/27/2012 19:30, vivo75@gmail.com wrote:

> Il 28/03/2012 01:18, Joshua Kinard ha scritto:
>> On 03/27/2012 11:42, Sven Vermeulen wrote:
>>
>>> Is it truely that much documentation? The handbook doesn't suggest separate
>>> /usr anymore (all example partitions use a single / file system), and the
>>> necessary changes to setup an initramfs are being prepared (cfr bugs related
>>> to bug #407959).
>>
>> Almost, see Code Listing 1.1:
>> http://www.gentoo.org/doc/en/security/security-handbook.xml?part=1&chap=4
>>
> while there, may really be a good idea to replace reiserfs with whatever else


Like resierfs4?  I think that's still capable of accessing data at plaid speed.

-- 
Joshua Kinard
Gentoo/MIPS
kumba@gentoo.org
4096R/D25D95E3 2011-03-28

"The past tempts us, the present confuses us, the future frightens us.  And
our lives slip away, moment by moment, lost in that vast, terrible in-between."

--Emperor Turhan, Centauri Republic


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 834 bytes --]

Re: [gentoo-project] Council meeting: Tuesday 3 April 2012, 19:00 UTC

[Ulrich Mueller]

>>>>> On Tue, 27 Mar 2012, Sven Vermeulen wrote:

>> 3. New udev and separate /usr partition (30 minutes)
>> 
>> See [4]: "Decide on whether a separate /usr is still a supported
>> configuration. If it is, newer udev can not be stabled and
>> alternatives should be investigated. If it isn't, a lot of
>> documentation will have to be updated. (And an alternative should
>> likely still be provided.)"

> Is it truely that much documentation? The handbook doesn't suggest
> separate /usr anymore (all example partitions use a single / file
> system), and the necessary changes to setup an initramfs are being
> prepared (cfr bugs related to bug #407959).

I've got a somewhat different impression, namely that it's still a
long way to go until the documentation will be sufficiently complete.

When updating my own systems (separate /usr partition, ext3 or ext4,
otherwise nothing special), I wasn't so happy with the status of the
documentation for the alternatives that are being discussed:

- dracut: I failed to find any documentation in our handbooks.
  Anyway, dracut looks like overkill for my simple usage case.

- genkernel: Easy to use, but it doesn't include e2fsck with the
  generated initramfs. Am I supposed to mount /usr without running
  fsck on it first? Once the partition is mounted (even read-only),
  e2fsck will refuse to run on it (so it also cannot be checked later
  in the boot sequence). Again, I was unable to find any documentation
  that would answer these questions.

So finally I ended up building my own initramfs, using information
from the wiki and from forums.

> We might want to provide an initramfs-guide to explain the concepts
> and the various methods for generating one (as well as debugging),
> but that too might not be the most difficult thing out there
> (there's still time for that, no)?

IMHO this has to be done before >=udev-180 can be stabilised.

Ulrich

Re: [gentoo-project] Council meeting: Tuesday 3 April 2012, 19:00 UTC

[Rich Freeman]

On Wed, Mar 28, 2012 at 5:15 PM, Ulrich Mueller <ulm@gentoo.org> wrote:
> - dracut: I failed to find any documentation in our handbooks.
>  Anyway, dracut looks like overkill for my simple usage case.
>

I dunno - dracut aims to be the ultimate generic automagic
plugin-supporting option, so I wouldn't dismiss it out of hand.  Its
biggest issue is that its features have overtaken the documentation,
and it is still pretty raw.

> So finally I ended up building my own initramfs, using information
> from the wiki and from forums.

I think you would have been better off with dracut, but as I've
blogged it isn't without its issues.  I might give genkernel a shot
again now that it sounds like it has better handling of
md/lvm/bind/etc.  If it can figure out my setup then I'll be
impressed.

My biggest issue with dracut is that for whatever reason it doesn't
detect and assemble my raid.  I ended up hacking together a module to
force it to run:
        mdadm -As
        lvm pvscan
        lvm vgscan
        lvm lvscan
        lvm vgchange -ay

I just can't get how if that works how the automagic functionality in
the script can't work.  The part that fails is mdadm bit, and it works
fine on full-auto mode (though it does have my mdadm.conf to look at).
 Even the kernel auto-assembles the raid fine, though that code is
disabled if you use an initramfs.

>
> IMHO this has to be done before >=udev-180 can be stabilised.
>

I agree completely.  The current documentation is too weak for stabilization.

Rich

Re: [gentoo-project] Council meeting: Tuesday 3 April 2012, 19:00 UTC

[Roy Bamford]

On 2012.03.28 22:46, Rich Freeman wrote:
> On Wed, Mar 28, 2012 at 5:15 PM, Ulrich Mueller <ulm@gentoo.org>
> wrote:
> > - dracut: I failed to find any documentation in our handbooks.
> >  Anyway, dracut looks like overkill for my simple usage case.
> >
> 
> I dunno - dracut aims to be the ultimate generic automagic
> plugin-supporting option, so I wouldn't dismiss it out of hand.  Its
> biggest issue is that its features have overtaken the documentation,
> and it is still pretty raw.
> 
[snip]
> 
> Rich
> 
> 

I really don't want any autoblackmagic involved in the booting of my 
systems.  If I don't understand it I can't fix it when (not if) it 
breaks. 

I have separate /usr everywhere and most places raid and lvm, so I 
already have initrds most places. I will reluctantly move to an initrd 
everywhere - but only if I can roll it by hand, so I can fix it by hand 
too.

-- 
Regards,

Roy Bamford
(Neddyseagoon) a member of
elections
gentoo-ops
forum-mods
trustees

Re: [gentoo-project] Council meeting: Tuesday 3 April 2012, 19:00 UTC

[William Hubbs]

[-- Attachment #1: Type: text/plain, Size: 549 bytes --]

On Wed, Mar 28, 2012 at 11:13:34PM +0100, Roy Bamford wrote:
> I really don't want any autoblackmagic involved in the booting of my 
> systems.  If I don't understand it I can't fix it when (not if) it 
> breaks. 
> 
> I have separate /usr everywhere and most places raid and lvm, so I 
> already have initrds most places. I will reluctantly move to an initrd 
> everywhere - but only if I can roll it by hand, so I can fix it by hand 
> too.

You could; there would be nothing stopping you from rolling your own
initramfs.

William


[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]