[gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Nikos Chantziaras]

On 28/09/13 01:33, Dale wrote:
> Bruce Hill wrote:
>> mingdao@workstation ~ $ eselect news read
>> 2013-09-27-initramfs-required
>>    Title                     Separate /usr on Linux requires initramfs
>>    Author                    William Hubbs <williamh@gentoo.org>
>>    Posted                    2013-09-27
>>    Revision                  1
>>
>> Linux systems which have / and /usr on separate file systems but do not
>> use an initramfs will not be supported starting on 01-Nov-2013.
>>  [...]
>
>
> I'm hoping that since I use eudev, I don't have to worry about this.  If
> I do, this could get interesting, again.

You do need to worry about this.  Actually, you always had to worry 
about this.  It's just that your specific configuration didn't blow up 
in a visible way.  You might had problems already in the past, just not 
apparent ones.  If you read the links posted in the announcement, you 
will see that the problem wasn't eudev or udev.  It's all the other 
software on your system.

eudev *cannot* fix that.

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Alon Bar-Lev]

On Sat, Sep 28, 2013 at 11:43 PM, Nikos Chantziaras <realnc@gmail.com> wrote:
> On 28/09/13 01:33, Dale wrote:
>>
>> Bruce Hill wrote:
>>>
>>> mingdao@workstation ~ $ eselect news read
>>> 2013-09-27-initramfs-required
>>>    Title                     Separate /usr on Linux requires initramfs
>>>    Author                    William Hubbs <williamh@gentoo.org>
>>>    Posted                    2013-09-27
>>>    Revision                  1
>>>
>>> Linux systems which have / and /usr on separate file systems but do not
>>> use an initramfs will not be supported starting on 01-Nov-2013.
>>>  [...]
>>
>>
>>
>> I'm hoping that since I use eudev, I don't have to worry about this.  If
>> I do, this could get interesting, again.
>
>
> You do need to worry about this.  Actually, you always had to worry about
> this.  It's just that your specific configuration didn't blow up in a
> visible way.  You might had problems already in the past, just not apparent
> ones.  If you read the links posted in the announcement, you will see that
> the problem wasn't eudev or udev.  It's all the other software on your
> system.
>
> eudev *cannot* fix that.
>

As far as I read, the problem is with bluetooth keyboards? and some
other devices and locales, which are minor for this decision of
removing supportability. Especially for servers and for most of
workstations. Most sane configuration can be supported with separate
/.

And of course there is the hidden systemd agenda, which is what I
suspect had more impact.

Regards,
Alon Bar-Lev.

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Alan McKinnon]

On 28/09/2013 22:58, Alon Bar-Lev wrote:
> As far as I read, the problem is with bluetooth keyboards? and some
> other devices and locales, which are minor for this decision of
> removing supportability. Especially for servers and for most of
> workstations. Most sane configuration can be supported with separate
> /.
> 
> And of course there is the hidden systemd agenda, which is what I
> suspect had more impact.

No, the problem is not bluetooth keyboards per se. That just happens to
be a convenient example of the kind of problem anticipated. There is a
tendency to use it as the only example, which reinforces the idea that
BT keyboards are problem to be solved.

The actual problem is better stated something like this:

In the early stages of user-land setup (around the time when udev is
getting it's act together), arbitrary code can run and that code can be
in any arbitrary place, but there is no guarantee that that code is even
accessible at the point when it is needed. The actual cause of this mess
is the lack of standards on where to put stuff on Linux systems, and it
forms a classic bootstrap problem.

There has only ever been one way around that problem - define an exact
entry point that is guaranteed to be in a specific state. For current
userland this effectively means that everything that has traditionally
been in bin, sbin and lib in / and /usr must be available as step 1.
Technically, you could include /var/lib/ and maybe even /opt in there.
but we can safely exclude those at this time as only a brain-dead moron
would ever put init-critical code there.

It's a fact of history that Linux packager and package devs have never
managed to make up their minds where to put stuff. Just have a look at
coreutils binaries - why are 60% of them in /usr? It's coreutils! And
core isn't in the name because of a whim.

So you have two choices: enforce a decent separation so that the problem
doesn't happen, or enforce that all binaries are in one place where we
can call it "the system". Every major OS out there does the latter, it's
only Linux that tolerates this free for all wild wild west approach of
stick anything anywhere and still expect it to work. Hint: it doesn't
work. Duct-tape and bubblegum don't actually hold stuff together, no
matter how much we try convince ourselves it does.

This should actually have been done when MAKEDEV was phased out in
favour of the now-defunct devfs, but it's never too late to fix design
flaws that date back 30 years or more.

So this brings us back to the essential technical problem that still
needs to be solved on your machines:

/usr needs to be available (and not only for BT keyboards) at the
earliest possible opportunity - this is a technical constraint. To
guarantee that, you need to either merge /usr with /, or use an
initramfs to guarantee that /usr is available before anything else
happens in userland.

It *really* is that simple. If you have a better solution than my last
two choices, then I am all ears.


-- 
Alan McKinnon
alan.mckinnon@gmail.com

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Volker Armin Hemmann]

Am 29.09.2013 00:36, schrieb Alan McKinnon:
> On 28/09/2013 22:58, Alon Bar-Lev wrote:
>> As far as I read, the problem is with bluetooth keyboards? and some
>> other devices and locales, which are minor for this decision of
>> removing supportability. Especially for servers and for most of
>> workstations. Most sane configuration can be supported with separate
>> /.
>>
>> And of course there is the hidden systemd agenda, which is what I
>> suspect had more impact.
> No, the problem is not bluetooth keyboards per se. That just happens to
> be a convenient example of the kind of problem anticipated. There is a
> tendency to use it as the only example, which reinforces the idea that
> BT keyboards are problem to be solved.
>
> The actual problem is better stated something like this:
>
> In the early stages of user-land setup (around the time when udev is
> getting it's act together), arbitrary code can run and that code can be
> in any arbitrary place, but there is no guarantee that that code is even
> accessible at the point when it is needed. The actual cause of this mess
> is the lack of standards on where to put stuff on Linux systems, and it
> forms a classic bootstrap problem.
>
> There has only ever been one way around that problem - define an exact
> entry point that is guaranteed to be in a specific state. For current
> userland this effectively means that everything that has traditionally
> been in bin, sbin and lib in / and /usr must be available as step 1.
> Technically, you could include /var/lib/ and maybe even /opt in there.
> but we can safely exclude those at this time as only a brain-dead moron
> would ever put init-critical code there.
>
> It's a fact of history that Linux packager and package devs have never
> managed to make up their minds where to put stuff. Just have a look at
> coreutils binaries - why are 60% of them in /usr? It's coreutils! And
> core isn't in the name because of a whim.
>
> So you have two choices: enforce a decent separation so that the problem
> doesn't happen, or enforce that all binaries are in one place where we
> can call it "the system". Every major OS out there does the latter, it's
> only Linux that tolerates this free for all wild wild west approach of
> stick anything anywhere and still expect it to work. Hint: it doesn't
> work. Duct-tape and bubblegum don't actually hold stuff together, no
> matter how much we try convince ourselves it does.
>
> This should actually have been done when MAKEDEV was phased out in
> favour of the now-defunct devfs, but it's never too late to fix design
> flaws that date back 30 years or more.
>
> So this brings us back to the essential technical problem that still
> needs to be solved on your machines:
>
> /usr needs to be available (and not only for BT keyboards) at the
> earliest possible opportunity - this is a technical constraint. To
> guarantee that, you need to either merge /usr with /, or use an
> initramfs to guarantee that /usr is available before anything else
> happens in userland.
>
> It *really* is that simple. If you have a better solution than my last
> two choices, then I am all ears.
>
>

the correct and simple solution would be to deprecate /usr and move
everything into / .

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[pk]

On 2013-09-29 01:23, Volker Armin Hemmann wrote:

> the correct and simple solution would be to deprecate /usr and move
> everything into / .

Install Windows and be done with it, I say.

Best regards

Peter K

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Dale]

pk wrote:
> On 2013-09-29 01:23, Volker Armin Hemmann wrote:
>
>> the correct and simple solution would be to deprecate /usr and move
>> everything into / .
> Install Windows and be done with it, I say.
>
> Best regards
>
> Peter K
>
>
>


Next, we'll have to have C: even tho we never had to have one before. 
ROFLMBO

Dale

:-)  :-) 

-- 
I am only responsible for what I said ... Not for what you understood or how you interpreted my words!

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Alan McKinnon]

On 29/09/2013 01:23, Volker Armin Hemmann wrote:
>> It *really* is that simple. If you have a better solution than my last
>> > two choices, then I am all ears.
>> >
>> >
> the correct and simple solution would be to deprecate /usr and move
> everything into / .
> 
> 

I did consider that, but gave up on the idea as not workable. Sure, it
would work great and did work very well for Android and MacOS, both
controlled environments.

But doing it gains you nothing really apart from a crap load of stuff
cluttering up /, thinks like local, games and share.

But hey, maybe we can go right back to the originsl and put /home where
it started: /usr/people

-- 
Alan McKinnon
alan.mckinnon@gmail.com

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Alan McKinnon]

On 29/09/2013 02:01, Dale wrote:
> pk wrote:
>> On 2013-09-29 01:23, Volker Armin Hemmann wrote:
>>
>>> the correct and simple solution would be to deprecate /usr and move
>>> everything into / .
>> Install Windows and be done with it, I say.
>>
>> Best regards
>>
>> Peter K
>>
>>
>>
> 
> 
> Next, we'll have to have C: even tho we never had to have one before. 
> ROFLMBO
> 
> Dale
> 
> :-)  :-) 
> 


You can have a C any time you want, all you need is a simple

s#/#C#g

-- 
Alan McKinnon
alan.mckinnon@gmail.com

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[pk]

On 2013-09-29 02:01, Dale wrote:

> Next, we'll have to have C: even tho we never had to have one before. 
> ROFLMBO

I would hesitate to laugh because that's where Linux is heading... And
Alan and other's are right in that it's not udevs problem per se; it's
all the half-desktop services[1]/applications that requires access to
the libs in /usr for some unknown reason. This will (eventually) affect
any operating system (even FreeBSD) that want to run things like, say,
Gnome. This is feature creep on steroids.

I just wish there was this simple system that would look like this:

boot loader -> operating system -> applications

...with a clear separation/well defined interfaces between them. Used to
think Linux was a good compromise but not anymore... What you have now
is something monstrous where application libs are part of the operating
system. Hence the requirement of no separate /usr. At least if you run
any of those things (like PAM - if some module require access to
PKCS#11, Kerberos, Consolekit etc.). Personally I wouldn't touch them...

In my opinion, this has gone way beyond what used to be called
"spaghetti code" and into what I would like to call "spaghetti system".

[1] Used to be called daemons but now people have adopted the Windows
name for it.

Best regards

Peter K

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Bruce Hill]

On Sat, Sep 28, 2013 at 07:01:56PM -0500, Dale wrote:
> 
> Next, we'll have to have C: even tho we never had to have one before. 
> ROFLMBO
> 
> Dale

We already have it, just we don't have to CAPITALIZE c:

mingdao@workstation ~ $ ls -l .wine/drive_c/
total 8
drwxr-xr-x  6 mingdao mingdao  107 May 16 08:46 Program Files
-rw-r--r--  1 mingdao mingdao  529 Nov  1  2012 teamviewer.html
drwxr-xr-x  4 mingdao mingdao   33 Nov  1  2012 users
drwxr-xr-x 14 mingdao mingdao 4096 Sep 20 11:41 windows

But, seriously; our Linux desktop systems are so far behind Windows it really
makes us look bad. We kick tail in the server market, but that's it.
-- 
Happy Penguin Computers               >')
126 Fenco Drive                       ( \
Tupelo, MS 38801                       ^^
support@happypenguincomputers.com
662-269-2706 662-205-6424
http://happypenguincomputers.com/

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

Don't top-post: http://en.wikipedia.org/wiki/Top_post#Top-posting

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Walter Dnes]

On Sun, Sep 29, 2013 at 12:36:43AM +0200, Alan McKinnon wrote

> The actual problem is better stated something like this:
> 
> In the early stages of user-land setup (around the time when udev is
> getting it's act together), arbitrary code can run and that code can be
> in any arbitrary place, but there is no guarantee that that code is even
> accessible at the point when it is needed. The actual cause of this mess
> is the lack of standards on where to put stuff on Linux systems, and it
> forms a classic bootstrap problem.
> 
> There has only ever been one way around that problem - define an exact
> entry point that is guaranteed to be in a specific state. For current
> userland this effectively means that everything that has traditionally
> been in bin, sbin and lib in / and /usr must be available as step 1.
> Technically, you could include /var/lib/ and maybe even /opt in there.
> but we can safely exclude those at this time as only a brain-dead moron
> would ever put init-critical code there.

  Separate /usr worked for many years, even with udev.  The question I
have is why is udev *NOW* monkeying around with a whole bunch of
additional stuff before mounting partitions?  If you have an NFS-mounted
/usr, I can see needing to have network services running first.  Ditto
for /usr being in an LVM or encrypted partition, you need LVM and/or
decryption running first.  There is no excuse for anything else breaking
a separate /usr.

  Then again, separate /usr isn't the first thing Kay Sievers has broken
since he took over udev, and I wouldn't be surprised if he one day "just
happens to break openrc"...

https://lkml.org/lkml/2012/10/2/303

> From	Linus Torvalds <>
> Date	Tue, 2 Oct 2012 09:33:03 -0700
> Subject Re: udev breakages - was: Re: Need of an ".async_probe()"
> type of callback at driver's core - Was: Re: [PATCH] [media] drxk:
> change it to use request_firmware_nowait()
	

> On Tue, Oct 2, 2012 at 6:03 AM, Mauro Carvalho Chehab
> <mchehab@redhat.com> wrote:
>
> > I basically tried a few different approaches, including deferred
> > probe(), as you suggested, and request_firmware_async(), as Kay
> > suggested.
>
> Stop this crazy. FIX UDEV ALREADY, DAMMIT.
>
> Who maintains udev these days? Is it Lennart/Kai, as part of systemd?
>
> Lennart/Kai, fix the udev regression already. Lennart was the one
> who brought up kernel ABI regressions at some conference, and if
> you now you have the *gall* to break udev in an incompatible manner
> that requires basically impossible kernel changes for the kernel to
> "fix" the udev interface, I don't know what to say.
> 
> "Two-faced lying weasel" would be the most polite thing I could say.
> But it almost certainly will involve a lot of cursing.
>
> > However, for 3.7 or 3.8, I think that the better is to revert
> > changeset 177bc7dade38b5 and to stop with udev's insanity of
> > requiring asynchronous firmware load during device driver
> > initialization. If udev's developers are not willing to do that,
> > we'll likely need to  add something at the drivers core to trick
> > udev for it to think that the modules got probed before the probe
> > actually happens.
> 
> The fact is, udev made new - and insane - rules that are simply
> *invalid*. Modern udev is broken, and needs to be fixed.
> 
> I don't know where the problem started in udev, but the report I
> saw was that udev175 was fine, and udev182 was broken, and would
> deadlock if module_init() did a request_firmware(). That kind of
> nested behavior is absolutely *required* to work, in order to not
> cause idiotic problems for the kernel for no good reason.
> 
> What kind of insane udev maintainership do we have? And can we fix it?

-- 
Walter Dnes <waltdnes@waltdnes.org>
I don't run "desktop environments"; I run useful applications

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[pk]

On 2013-09-29 08:06, Walter Dnes wrote:

>> What kind of insane udev maintainership do we have? And can we fix it?

By starting from scratch and putting it in the kernel (which will stop
people from being too "creative" as well, since Linus will not allow
things to break so easily). The BSDs, MacOS and Plan 9 kernels can do
it[1], why not Linux? Well, one can wish at least... :-)

[1] https://en.wikipedia.org/wiki/Devfs#Implementations

Best regards

Peter K

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 539 bytes --]

On Sun, 29 Sep 2013 02:06:34 -0400, Walter Dnes wrote:

> for /usr being in an LVM or encrypted partition, you need LVM and/or
> decryption running first.

Why would you want /usr encrypted but not /? There is nothing private
in /usr, but /etc/ contains password files.

I have used a separate usr in the past to do it the other way round,
encrypted / but unencrypted /usr (to lower processor usage on a netbook)
but that requires an initramfs anyway.


-- 
Neil Bothwick

Justify my text? I'm sorry but it has no excuse.

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

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Volker Armin Hemmann]

Am 29.09.2013 02:08, schrieb Alan McKinnon:
> On 29/09/2013 01:23, Volker Armin Hemmann wrote:
>>> It *really* is that simple. If you have a better solution than my last
>>>> two choices, then I am all ears.
>>>>
>>>>
>> the correct and simple solution would be to deprecate /usr and move
>> everything into / .
>>
>>
> I did consider that, but gave up on the idea as not workable. Sure, it
> would work great and did work very well for Android and MacOS, both
> controlled environments.
>
> But doing it gains you nothing really apart from a crap load of stuff
> cluttering up /, thinks like local, games and share.
>
> But hey, maybe we can go right back to the originsl and put /home where
> it started: /usr/people
>
and a cluttered / is worse than a non-existant / and a cluttered /usr?

Because we are just moving in that direction.

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Volker Armin Hemmann]

Am 29.09.2013 01:31, schrieb pk:
> On 2013-09-29 01:23, Volker Armin Hemmann wrote:
>
>> the correct and simple solution would be to deprecate /usr and move
>> everything into / .
> Install Windows and be done with it, I say.
>
> Best regards
>
> Peter K
>
>
> .
>
look at history, think and retry.

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[pk]

On 2013-09-29 12:59, Volker Armin Hemmann wrote:

> look at history, think and retry.

That's just what I did. Read and retry.

Best regards

Peter K

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Volker Armin Hemmann]

Am 29.09.2013 17:24, schrieb pk:
> On 2013-09-29 12:59, Volker Armin Hemmann wrote:
>
>> look at history, think and retry.
> That's just what I did. Read and retry.
>
> Best regards
>
> Peter K
>
>
> .
>
I did, your mail did not make any more sense at all.

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Dale]

Volker Armin Hemmann wrote:
> Am 29.09.2013 17:24, schrieb pk:
>> On 2013-09-29 12:59, Volker Armin Hemmann wrote:
>>
>>> look at history, think and retry.
>> That's just what I did. Read and retry.
>>
>> Best regards
>>
>> Peter K
>>
>>
>> .
>>
> I did, your mail did not make any more sense at all.
>
>

That could be the problem then couldn't it? 

Dale

:-)  :-) 

-- 
I am only responsible for what I said ... Not for what you understood or how you interpreted my words!

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[pk]

On 2013-09-29 18:36, Dale wrote:

> That could be the problem then couldn't it? 

Indeed. :-)

Best regards

Peter K

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Tanstaafl]

On 2013-09-28 6:36 PM, Alan McKinnon <alan.mckinnon@gmail.com> wrote:
> So this brings us back to the essential technical problem that still
> needs to be solved on your machines:
>
> /usr needs to be available (and not only for BT keyboards) at the
> earliest possible opportunity - this is a technical constraint. To
> guarantee that, you need to either merge /usr with /, or use an
> initramfs to guarantee that /usr is available before anything else
> happens in userland.
>
> It*really*  is that simple. If you have a better solution than my last
> two choices, then I am all ears.

Ok, and if this is all true, I can accept it.

But...

> Technically, you could include /var/lib/ and maybe even /opt in there.
> but we can safely exclude those at this time as only a brain-dead moron
> would ever put init-critical code there.

I also have /var on a separate (LVM) partition. What I'm AFRAID of, is 
that some 'brain-dead moron' will, sometime in the future, arbitrarily 
decide that having a separate /var will *also* require an initramfs 
because some *other* brain-dead moron (who happens to have enough clout 
to shove their garbage down our throats)... then what is next /home?

It seems to me like the more likely case is that someone somewhere wants 
to require BOTH systemd AND an initramfs in ALL cases, and this is just 
the first step in that progression.

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[William Hubbs]

[-- Attachment #1: Type: text/plain, Size: 1266 bytes --]

On Sun, Sep 29, 2013 at 01:55:49PM -0400, Tanstaafl wrote:
> On 2013-09-28 6:36 PM, Alan McKinnon <alan.mckinnon@gmail.com> wrote:
> > So this brings us back to the essential technical problem that still
> > needs to be solved on your machines:
> >
> > /usr needs to be available (and not only for BT keyboards) at the
> > earliest possible opportunity - this is a technical constraint. To
> > guarantee that, you need to either merge /usr with /, or use an
> > initramfs to guarantee that /usr is available before anything else
> > happens in userland.
> >
> > It*really*  is that simple. If you have a better solution than my last
> > two choices, then I am all ears.
> 
> Ok, and if this is all true, I can accept it.

Alan, this is a very good summary of the issues involved. Everyone on
the list should go and read flameeyes' blog post then this summary.

Tanstaaf,

I am the OpenRC author/maintainer and a member of base-system. I can
tell you that we are not discussing forcing systemd on everyone in
Gentoo Linux as a default init system. I can also tell you that I am not
aware of the Gentoo systemd team discussing this. Even if they were, a
distro-wide change like this would have to be brought before the
Council.

William


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

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Tanstaafl]

On 2013-09-29 2:55 PM, William Hubbs <williamh@gentoo.org> wrote:
> I am the OpenRC author/maintainer and a member of base-system. I can
> tell you that we are not discussing forcing systemd on everyone in
> Gentoo Linux as a default init system. I can also tell you that I am not
> aware of the Gentoo systemd team discussing this. Even if they were, a
> distro-wide change like this would have to be brought before the
> Council.

Ok, good enough for me until other evidence comes along to cast doubt as 
to the truthfulness or sincerity of your statement.

Thanks William...

Now to try to get up enough nerve to attempt to merge my /usr (currently 
on LVM partition) into my / (does have enough room, and will leave me 
with a 19GB / partition with about 5GB free).

Anyone see a problem with that (only 5GB free on my / after the /usr merge)?

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Alan McKinnon]

On 29/09/2013 20:55, William Hubbs wrote:
> On Sun, Sep 29, 2013 at 01:55:49PM -0400, Tanstaafl wrote:
>> On 2013-09-28 6:36 PM, Alan McKinnon <alan.mckinnon@gmail.com> wrote:
>>> So this brings us back to the essential technical problem that still
>>> needs to be solved on your machines:
>>>
>>> /usr needs to be available (and not only for BT keyboards) at the
>>> earliest possible opportunity - this is a technical constraint. To
>>> guarantee that, you need to either merge /usr with /, or use an
>>> initramfs to guarantee that /usr is available before anything else
>>> happens in userland.
>>>
>>> It*really*  is that simple. If you have a better solution than my last
>>> two choices, then I am all ears.
>>
>> Ok, and if this is all true, I can accept it.
> 
> Alan, this is a very good summary of the issues involved. Everyone on
> the list should go and read flameeyes' blog post then this summary.

Thanks William.

It really was an off-the cuff description done to answer a user's
question. I'm glad to hear it communicated what I intended.



> 
> Tanstaaf,
> 
> I am the OpenRC author/maintainer and a member of base-system. I can
> tell you that we are not discussing forcing systemd on everyone in
> Gentoo Linux as a default init system. I can also tell you that I am not
> aware of the Gentoo systemd team discussing this. Even if they were, a
> distro-wide change like this would have to be brought before the
> Council.
> 
> William
> 


-- 
Alan McKinnon
alan.mckinnon@gmail.com

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Tanstaafl]

Weird - I thought I replied to this a while ago (I know I started one), 
but it disappeared, and is not in my Sent folder and it never made it to 
the list...

On 2013-09-29 2:55 PM, William Hubbs <williamh@gentoo.org> wrote:
> I am the OpenRC author/maintainer and a member of base-system. I can
> tell you that we are not discussing forcing systemd on everyone in
> Gentoo Linux as a default init system. I can also tell you that I am not
> aware of the Gentoo systemd team discussing this. Even if they were, a
> distro-wide change like this would have to be brought before the
> Council.

Thanks very much for this William. I will take you at your word and will 
stop worrying about the whole systemd thing (unless/until evidence 
warrants revisiting it)...

So, now I just have to get up the nerve to attempt the merging of my LVM 
based /usr into my / so I don't have to worry about an initramfs.

There are no technical reasons it shouldn't work - my / is 19G, with 
18GB free right now. My /usr currently takes up 13GB, so merging should 
leave mw with 5GB free...

Does anyone see an issue with a 19GB / with merged /usr and only 5GB 
free? Was I correct in my statement to Dale that there is nothing used 
by or stored in /usr that could consume that last 5GB and crash my 
server (ie, like a runaway log can fill up /var)?

Thanks again...

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Alan McKinnon]

On 29/09/2013 22:51, Tanstaafl wrote:
> Weird - I thought I replied to this a while ago (I know I started one),
> but it disappeared, and is not in my Sent folder and it never made it to
> the list...
> 
> On 2013-09-29 2:55 PM, William Hubbs <williamh@gentoo.org> wrote:
>> I am the OpenRC author/maintainer and a member of base-system. I can
>> tell you that we are not discussing forcing systemd on everyone in
>> Gentoo Linux as a default init system. I can also tell you that I am not
>> aware of the Gentoo systemd team discussing this. Even if they were, a
>> distro-wide change like this would have to be brought before the
>> Council.
> 
> Thanks very much for this William. I will take you at your word and will
> stop worrying about the whole systemd thing (unless/until evidence
> warrants revisiting it)...
> 
> So, now I just have to get up the nerve to attempt the merging of my LVM
> based /usr into my / so I don't have to worry about an initramfs.
> 
> There are no technical reasons it shouldn't work - my / is 19G, with
> 18GB free right now. My /usr currently takes up 13GB, so merging should
> leave mw with 5GB free...
> 
> Does anyone see an issue with a 19GB / with merged /usr and only 5GB
> free? Was I correct in my statement to Dale that there is nothing used
> by or stored in /usr that could consume that last 5GB and crash my
> server (ie, like a runaway log can fill up /var)?
> 
> Thanks again...
> 

Correct on all counts. This laptop runs KDE, here's my breakdown:

# du -sh /usr
13G     /usr

# du -sh /usr/*
12K     /usr/INSTALL
104K    /usr/Licenses_for_Third-Party_Components.txt
426M    /usr/bin
12M     /usr/gnu-classpath-0.98
460M    /usr/include
0       /usr/lib
525M    /usr/lib32
2.8G    /usr/lib64
134M    /usr/libexec
512K    /usr/local
38M     /usr/sbin
3.6G    /usr/share
4.9G    /usr/src
0       /usr/tmp
11M     /usr/x86_64-pc-linux-gnu

Those numbers are not likely to change much with time, with one exception:

/usr/src

That can get real big real quick if you don't clean up kernel sources
often. Ideally, you'd make that a suitably sized LV and mount it seperately.

The other space consumer is /usr/share with it's many documentation
files. But those too tend to be stable once you have everything
installed. 5G free out of 19G is ~75% space in use which is perfectly
acceptable for this case.

Regular monitoring of the state of your machines will tell you if space
usage increases so you can investigate and deal with it timeously.

I assume you long since moved portage and it's storage directories out
of /usr into /var?



-- 
Alan McKinnon
alan.mckinnon@gmail.com

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Tanstaafl]

On 2013-09-29 5:15 PM, Alan McKinnon <alan.mckinnon@gmail.com> wrote:
> Those numbers are not likely to change much with time, with one exception:
>
> /usr/src
>
> That can get real big real quick if you don't clean up kernel sources
> often. Ideally, you'd make that a suitably sized LV and mount it seperately.

Yeah, I always keep 2 or 3 known good kernels, and clean out the old 
stuff, so no worries there.

> The other space consumer is /usr/share with it's many documentation
> files. But those too tend to be stable once you have everything
> installed. 5G free out of 19G is ~75% space in use which is perfectly
> acceptable for this case.
>
> Regular monitoring of the state of your machines will tell you if space
> usage increases so you can investigate and deal with it timeously.
>
> I assume you long since moved portage and it's storage directories out
> of /usr into /var?

Hmmm... No, I never did that myself...

Wow...

moria : Sun Sep 29, 18:19:01 : ~
  # du -sh /usr/*
85M     /usr/bin
131M    /usr/include
0       /usr/lib
11M     /usr/lib32
530M    /usr/lib64
51M     /usr/libexec
15M     /usr/local
7.8G    /usr/portage
21M     /usr/sbin
509M    /usr/share
3.9G    /usr/src
0       /usr/tmp
7.0M    /usr/x86_64-pc-linux-gnu
moria : Sun Sep 29, 18:26:30 : ~
  #

Is this the official gentoo way now? Will a new/fresh virgin install 
have /var/portage instead of /usr/portage?

I can eliminate almost 8GB by moving portage and its storage directories...

I don't recall seeing a news item about that...

But... is /usr/portage the default/recommended location? If so, then I 
don't think I want to move it - I generally never change defaults unless 
there is a very good reason to do so.

But, is there some official gentoo docs online explaining how to do this?

Something more to think about...

Also - is there any kind of maintenance I shoudl be doing on 
/usr/portage to clean old cruft out? Or does portage maintain it already.

:)

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Volker Armin Hemmann]

Am 30.09.2013 00:53, schrieb Tanstaafl:
> On 2013-09-29 5:15 PM, Alan McKinnon <alan.mckinnon@gmail.com> wrote:
>> Those numbers are not likely to change much with time, with one
>> exception:
>>
>> /usr/src
>>
>> That can get real big real quick if you don't clean up kernel sources
>> often. Ideally, you'd make that a suitably sized LV and mount it
>> seperately.
>
> Yeah, I always keep 2 or 3 known good kernels, and clean out the old
> stuff, so no worries there.
>
>> The other space consumer is /usr/share with it's many documentation
>> files. But those too tend to be stable once you have everything
>> installed. 5G free out of 19G is ~75% space in use which is perfectly
>> acceptable for this case.
>>
>> Regular monitoring of the state of your machines will tell you if space
>> usage increases so you can investigate and deal with it timeously.
>>
>> I assume you long since moved portage and it's storage directories out
>> of /usr into /var?
>
> Hmmm... No, I never did that myself...
>
> Wow...
>
> moria : Sun Sep 29, 18:19:01 : ~
>  # du -sh /usr/*
> 85M     /usr/bin
> 131M    /usr/include
> 0       /usr/lib
> 11M     /usr/lib32
> 530M    /usr/lib64
> 51M     /usr/libexec
> 15M     /usr/local
> 7.8G    /usr/portage
> 21M     /usr/sbin
> 509M    /usr/share
> 3.9G    /usr/src
> 0       /usr/tmp
> 7.0M    /usr/x86_64-pc-linux-gnu
> moria : Sun Sep 29, 18:26:30 : ~
>  #
>
> Is this the official gentoo way now? Will a new/fresh virgin install
> have /var/portage instead of /usr/portage?
>
> I can eliminate almost 8GB by moving portage and its storage
> directories...
>
> I don't recall seeing a news item about that...
>
> But... is /usr/portage the default/recommended location? If so, then I
> don't think I want to move it - I generally never change defaults
> unless there is a very good reason to do so.
>
> But, is there some official gentoo docs online explaining how to do this?
>
> Something more to think about...
>
> Also - is there any kind of maintenance I shoudl be doing on
> /usr/portage to clean old cruft out? Or does portage maintain it already.
>
> :)
>
>

df -h        
Dateisystem    Größe Benutzt Verf. Verw% Eingehängt auf
/dev/root        59G     33G   24G   58% /
devtmpfs        7,8G       0  7,8G    0% /dev
tmpfs           1,6G    712K  1,6G    1% /run
shm             7,8G    1,1M  7,8G    1% /dev/shm
cgroup_root      10M       0   10M    0% /sys/fs/cgroup
/dev/sda1       197M     17M  181M    9% /boot/efi
/dev/sde1       110G     82G   23G   79% /home/energyman
tmpfs           1,0G    3,4M 1021M    1% /tmp
zfstank/data    3,6T    1,9T  1,8T   52% /mnt/data
zfstank/var     100G     16G   85G   16% /var
zfstank         1,8T    256K  1,8T    1% /zfstank

and I put PORTDIR into /var ages ago. I hate 'moving targets' like
PORTDIR in a static place like /usr.
7,8G    /var/portage
6,5G    /var/packages

but seriously, if seperate /usr is so important for you - running
genkernel really IS easy...

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Daniel Campbell]

[-- Attachment #1: Type: text/plain, Size: 2459 bytes --]

On 09/29/2013 01:55 PM, William Hubbs wrote:
> On Sun, Sep 29, 2013 at 01:55:49PM -0400, Tanstaafl wrote:
>> On 2013-09-28 6:36 PM, Alan McKinnon <alan.mckinnon@gmail.com> wrote:
>>> So this brings us back to the essential technical problem that still
>>> needs to be solved on your machines:
>>>
>>> /usr needs to be available (and not only for BT keyboards) at the
>>> earliest possible opportunity - this is a technical constraint. To
>>> guarantee that, you need to either merge /usr with /, or use an
>>> initramfs to guarantee that /usr is available before anything else
>>> happens in userland.
>>>
>>> It*really*  is that simple. If you have a better solution than my last
>>> two choices, then I am all ears.
>>
>> Ok, and if this is all true, I can accept it.
> 
> Alan, this is a very good summary of the issues involved. Everyone on
> the list should go and read flameeyes' blog post then this summary.
> 
> Tanstaaf,
> 
> I am the OpenRC author/maintainer and a member of base-system. I can
> tell you that we are not discussing forcing systemd on everyone in
> Gentoo Linux as a default init system. I can also tell you that I am not
> aware of the Gentoo systemd team discussing this. Even if they were, a
> distro-wide change like this would have to be brought before the
> Council.
> 
> William
> 

I understand Gentoo has a much more structured way of making decisions
like systemd, but perhaps you aren't the best person to assuage fears. I
say this not because of anything you do, but your position. Arch Linux
used their sysvinit maintainer to calm fears of users before their
switch to systemd. I'm not saying that you are trying to do this at all,
but rather being OpenRC maintainer doesn't add much in the way of
credibility of your stance. Everything else (the lack of discussion on
it, the fact that the Council would have to vote on it) are much better
logical support for systemd not being forced.

I'm not sure if you knew about what happened with Arch, so I just
figured I'd point it out. I and others who switched from Arch to Gentoo
over the systemd debacle still remember the false promises (from the
sysvinit maintainer) that systemd won't be forced, when it was. So one's
position can't really be trusted, regardless of how much I and others
appreciate the work that goes into OpenRC.

No offense is intended, by the way. Just adding some context. I hope you
understand.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 555 bytes --]

[gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[»Q«]

On Sun, 29 Sep 2013 09:25:05 +0100
Mick <michaelkintzios@gmail.com> wrote:

> On Sunday 29 Sep 2013 06:29:37 Walter Dnes wrote:
> > On Sat, Sep 28, 2013 at 06:09:40PM -0500, Dale wrote
> >   
> > > Most likely, I'll install Kubuntu to start.  Then I may roam
> > > around and test other distros until I find one I like.  Thing is,
> > > I already have a starting point.  
> > 
> >   I'm already looking.
> > http://forums.funtoo.org/viewtopic.php?id=2265 and they also
> > dislike systemd.  I think I could get to like it.  See also
> > http://bugs.funtoo.org/browse/FL-34  
> 
> Very interesting!  This looks as a logical way to put udev back in
> its userspace box and stop it breaking the OS, or did I understand it
> incorrectly?

Funtoo is using mdev.  drobbins plans to make make GNOME 3.8+ work
without systemd/udev as well, but so far he's been mum about how.

[gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[»Q«]

On Sun, 29 Sep 2013 10:39:35 -0500
Dale <rdalek1967@gmail.com> wrote:

> I think I'll update that Kubuntu disk right quick while I am thinking
> about it.  Fall back plan just in case.  ;-)

Make sure you notify the Kubuntu mailing list of your contingency plans
in case Kubuntu's init thingy gives you trouble.  ;)

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Alan McKinnon]

On 30/09/2013 00:53, Tanstaafl wrote:
> On 2013-09-29 5:15 PM, Alan McKinnon <alan.mckinnon@gmail.com> wrote:
>> Those numbers are not likely to change much with time, with one
>> exception:
>>
>> /usr/src
>>
>> That can get real big real quick if you don't clean up kernel sources
>> often. Ideally, you'd make that a suitably sized LV and mount it
>> seperately.
> 
> Yeah, I always keep 2 or 3 known good kernels, and clean out the old
> stuff, so no worries there.
> 
>> The other space consumer is /usr/share with it's many documentation
>> files. But those too tend to be stable once you have everything
>> installed. 5G free out of 19G is ~75% space in use which is perfectly
>> acceptable for this case.
>>
>> Regular monitoring of the state of your machines will tell you if space
>> usage increases so you can investigate and deal with it timeously.
>>
>> I assume you long since moved portage and it's storage directories out
>> of /usr into /var?
> 
> Hmmm... No, I never did that myself...
> 
> Wow...
> 
> moria : Sun Sep 29, 18:19:01 : ~
>  # du -sh /usr/*
> 85M     /usr/bin
> 131M    /usr/include
> 0       /usr/lib
> 11M     /usr/lib32
> 530M    /usr/lib64
> 51M     /usr/libexec
> 15M     /usr/local
> 7.8G    /usr/portage
> 21M     /usr/sbin
> 509M    /usr/share
> 3.9G    /usr/src
> 0       /usr/tmp
> 7.0M    /usr/x86_64-pc-linux-gnu
> moria : Sun Sep 29, 18:26:30 : ~
>  #

Apart from portage and src that all looks totally normal and unlikely to
vary much over time.



> Is this the official gentoo way now? Will a new/fresh virgin install
> have /var/portage instead of /usr/portage?

The new instaled default is to put all of portage on /var, whilst still
supporting old installs on /usr. This is no big deal in code, as it's
really just a string containing a base path


> I can eliminate almost 8GB by moving portage and its storage directories...

Or move them onto a dedictaed LV. This is a case where a different mount
point makes a lot of sense - we're all aware just how unique the tree is
in terms of fs performance - thousands of small files mostly smaller
than 2k in hundreds of directories. It's quite different to everything
else on /usr or even /var.

Same with distfiles, that too can move anywhere you want it to be, just
adjust one setting in make.conf

> I don't recall seeing a news item about that...

IIRC it wasn't a news item as such. Perhaps it was an elog from portage
itself.


> 
> But... is /usr/portage the default/recommended location? If so, then I
> don't think I want to move it - I generally never change defaults unless
> there is a very good reason to do so.

It's /var/portage for new installs. If you want it to be somewhere else,
just move it and adjust make.conf


> 
> But, is there some official gentoo docs online explaining how to do this?
> 
> Something more to think about...
> 
> Also - is there any kind of maintenance I shoudl be doing on
> /usr/portage to clean old cruft out? Or does portage maintain it already.

rsync takes care of all that.
You have eclean to keep distfiles tidy
binpkgs you need to clean up on your own, as portage has no way of
knowing what you want to keep. And local overlays fall in the same category


-- 
Alan McKinnon
alan.mckinnon@gmail.com

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Dale]

»Q« wrote:
> On Sun, 29 Sep 2013 10:39:35 -0500
> Dale <rdalek1967@gmail.com> wrote:
>
>> I think I'll update that Kubuntu disk right quick while I am thinking
>> about it.  Fall back plan just in case.  ;-)
> Make sure you notify the Kubuntu mailing list of your contingency plans
> in case Kubuntu's init thingy gives you trouble.  ;)
>
>
>

Real simple, reinstall.  It takes a very short time compared to Gentoo. 
I used to install Mandrake in about 30 minutes and that was a complete
install on much slower hard drives and CD readers. 

I got that covered.

Dale

:-)  :-) 

-- 
I am only responsible for what I said ... Not for what you understood or how you interpreted my words!

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Volker Armin Hemmann]

Am 30.09.2013 11:00, schrieb Alan McKinnon:
> On 30/09/2013 00:53, Tanstaafl wrote:
>> On 2013-09-29 5:15 PM, Alan McKinnon <alan.mckinnon@gmail.com> wrote:
>>> Those numbers are not likely to change much with time, with one
>>> exception:
>>>
>>> /usr/src
>>>
>>> That can get real big real quick if you don't clean up kernel sources
>>> often. Ideally, you'd make that a suitably sized LV and mount it
>>> seperately.
>> Yeah, I always keep 2 or 3 known good kernels, and clean out the old
>> stuff, so no worries there.
>>
>>> The other space consumer is /usr/share with it's many documentation
>>> files. But those too tend to be stable once you have everything
>>> installed. 5G free out of 19G is ~75% space in use which is perfectly
>>> acceptable for this case.
>>>
>>> Regular monitoring of the state of your machines will tell you if space
>>> usage increases so you can investigate and deal with it timeously.
>>>
>>> I assume you long since moved portage and it's storage directories out
>>> of /usr into /var?
>> Hmmm... No, I never did that myself...
>>
>> Wow...
>>
>> moria : Sun Sep 29, 18:19:01 : ~
>>  # du -sh /usr/*
>> 85M     /usr/bin
>> 131M    /usr/include
>> 0       /usr/lib
>> 11M     /usr/lib32
>> 530M    /usr/lib64
>> 51M     /usr/libexec
>> 15M     /usr/local
>> 7.8G    /usr/portage
>> 21M     /usr/sbin
>> 509M    /usr/share
>> 3.9G    /usr/src
>> 0       /usr/tmp
>> 7.0M    /usr/x86_64-pc-linux-gnu
>> moria : Sun Sep 29, 18:26:30 : ~
>>  #
> Apart from portage and src that all looks totally normal and unlikely to
> vary much over time.
>
>
>
>> Is this the official gentoo way now? Will a new/fresh virgin install
>> have /var/portage instead of /usr/portage?
> The new instaled default is to put all of portage on /var, whilst still
> supporting old installs on /usr. This is no big deal in code, as it's
> really just a string containing a base path
>
>
>> I can eliminate almost 8GB by moving portage and its storage directories...
> Or move them onto a dedictaed LV. This is a case where a different mount
> point makes a lot of sense - we're all aware just how unique the tree is
> in terms of fs performance - thousands of small files mostly smaller
> than 2k in hundreds of directories. It's quite different to everything
> else on /usr or even /var.
>
> Same with distfiles, that too can move anywhere you want it to be, just
> adjust one setting in make.conf
>
>> I don't recall seeing a news item about that...
> IIRC it wasn't a news item as such. Perhaps it was an elog from portage
> itself.
>
>
>> But... is /usr/portage the default/recommended location? If so, then I
>> don't think I want to move it - I generally never change defaults unless
>> there is a very good reason to do so.
> It's /var/portage for new installs. If you want it to be somewhere else,
> just move it and adjust make.conf
>
>
really? so when I moved PORTDIR to /var/portage I was ahead of the rest?
Wow...

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Alan McKinnon]

On 30/09/2013 19:25, Volker Armin Hemmann wrote:
>>> But... is /usr/portage the default/recommended location? If so, then I
>>> >> don't think I want to move it - I generally never change defaults unless
>>> >> there is a very good reason to do so.
>> > It's /var/portage for new installs. If you want it to be somewhere else,
>> > just move it and adjust make.conf
>> >
>> >
> really? so when I moved PORTDIR to /var/portage I was ahead of the rest?
> Wow...
> 


You were ahead of me for sure :-)

I clearly remember one day long long ago you ranted and raved about how
a huge chunk of /usr was write-often...

... so I fiddled with mine to make it work on /var too and was very
happy with it.



-- 
Alan McKinnon
alan.mckinnon@gmail.com

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Mick]

[-- Attachment #1: Type: Text/Plain, Size: 904 bytes --]

On Monday 30 Sep 2013 20:14:44 Alan McKinnon wrote:
> On 30/09/2013 19:25, Volker Armin Hemmann wrote:
> >>> But... is /usr/portage the default/recommended location? If so, then I
> >>> 
> >>> >> don't think I want to move it - I generally never change defaults
> >>> >> unless there is a very good reason to do so.
> >> > 
> >> > It's /var/portage for new installs. If you want it to be somewhere
> >> > else, just move it and adjust make.conf
> > 
> > really? so when I moved PORTDIR to /var/portage I was ahead of the rest?
> > Wow...
> 
> You were ahead of me for sure :-)
> 
> I clearly remember one day long long ago you ranted and raved about how
> a huge chunk of /usr was write-often...
> 
> ... so I fiddled with mine to make it work on /var too and was very
> happy with it.


There's no reason to move /usr/portage to /  It can stay in your LVM.

-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 490 bytes --]

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 799 bytes --]

On Mon, 30 Sep 2013 23:05:29 +0100, Mick wrote:

> > > really? so when I moved PORTDIR to /var/portage I was ahead of the
> > > rest? Wow...  
> > 
> > You were ahead of me for sure :-)
> > 
> > I clearly remember one day long long ago you ranted and raved about
> > how a huge chunk of /usr was write-often...
> > 
> > ... so I fiddled with mine to make it work on /var too and was very
> > happy with it.  
> 
> 
> There's no reason to move /usr/portage to /  It can stay in your LVM.

This isn't about moving it to /, it's about moving it to /var, which is a
far more logical location for the portage tree. /usr is for static system
files, /var is for variable data.


-- 
Neil Bothwick

What's the difference between ignorance and apathy?
I don't know and I don't care

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

[gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Jonathan Callen]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

On 09/30/2013 06:22 AM, Neil Bothwick wrote:
> On Mon, 30 Sep 2013 12:01:27 +0200, Hinnerk van Bruinehsen wrote:
> 
>>> mount /usr -o remount,ro mkdir /newusr rsync -a /usr/ /new/usr/ Comment out /usr line in
>>> /etc/fstab mv /usr /oldusr mv /newusr /usr reboot rmdir /oldusr
>>> 
>>> What you do with the old partition is up to you. In this case the discussion was about /usr
>>> on LVM, so you just delete it and allocate the space elsewhere when needed.
> 
>> You can even leave out the step of creating a new directory and moving it later if you
>> bind-mount you rootfs somewhere, e.g. /mnt/gentoo.
> 
> Good point.
> 
>> You may want to add some parameters to the call to rsync, though (e.g. those that preserve
>> permissions, xattrs (especially for SELinux or XT-PaX) and owner/group (should be -pogX),
> 
> -a covers most if not all of those.
> 
>> possibly -x aswell (if you have other filesystems under /usr (e.g. a discrete FS for the
>> portage tree).
> 
> Another good point, one of those things you think of immediately after hitting Send :(
> 
> 

Specifically, I would use -axAHX (-rlptgoD are implied by -a, but -HAX are not).

- From rsync(1):

- -a	archive mode; equals -rlptgoD (no -H,-A,-X)
- -r	recurse into directories
- -l	copy symlinks as symlinks
- -p	preserve permissions
- -t	preserve modification times
- -g	preserve group
- -o	preserve owner
- -D	preserve device files and special files
- -H	preserve hard links
- -A	preserve ACLs (implies -p)
- -X	preserve extended attributes
- -x	don't cross filesystem boundaries

- -- 
Jonathan Callen
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.21 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/
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=ef8P
-----END PGP SIGNATURE-----

[gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[James]

Gregory Shearman <zekeyg <at> gmail.com> writes:


> b) The important reason I need an initramfs is that I have my root
> filesystems on LVM partitions (except for my ARM servers). 

Hello Gregory,

Please tell me, as much as you are confortable  with, 
about your ARM servers....


Running Gentoo?  Running Embedded Gentoo?  Which kernels? 
HDD ? File Systems? Configurations, Grub 2? LVM, RAID ?
Typical usage?


What install docs did you follow?
Any suggestions on setting up ARM servers, cluster,
and such are most welcome.

etc etc etc.
curiously,
James

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Gregory Shearman]

In linux.gentoo.user, James wrote:
> Gregory Shearman <zekeyg <at> gmail.com> writes:
>
>
>> b) The important reason I need an initramfs is that I have my root
>> filesystems on LVM partitions (except for my ARM servers). 
>
> Hello Gregory,
>
> Please tell me, as much as you are confortable  with, 
> about your ARM servers....

I'm running 2 servers at the moment. They are very low power and they
mainly serve my home network. One is a Marvell Sheevaplug (single core
1.2GHZ 512MB memory) and has been running reliably for many years. The
other is a Texas Instruments Pandaboard (2 core Cortex A9 Processor -
1Gb memory) .  I've only had the Panda since October last year and it is
also a very reliable server (with added GUI HDMI benefits!).

> Running Gentoo?  Running Embedded Gentoo?  Which kernels? 
> HDD ? File Systems? Configurations, Grub 2? LVM, RAID ?

Both servers are running Gentoo Stable... therefore current kernels (for
their architecture). Both have external HDD attached via USB.

File systems: root filesystem is on an SDHC card (2nd partition). Other
filesystems (except for the boot partition) are all on LVM. I have
/usr/src, /usr/portage, /usr/portage/distfiles is a symlink to
/var/www/localhost/gentoo/distfiles (another filesystem). I also have
/var/tmp/portage on a separate filesystem and I also run a postgresql
database server which also has its own partition on
/var/lib/postgresql/<version>. Both servers have the same setup as I'm
currently in the process of replacing the sheevaplug with the panda.

Grub? There's no such thing on ARM machines. The kernel or uImage looks
for the first partition on the configured root device (SDHC on my
systems) the first partition MUST be VFAT (unfortunately) and it
contains the u-boot bootloader and the kernel (uImage). Kernels are
built the same way as x86 kernels except you do "make uImage" instead of
"make bzImage".

LVM? All the above filesystems, except the root partition and the boot
partition are LVM volumes. Filesystems are mostly Ext4 (very
conventional).

RAID? Nope.

> Typical usage?

Print server, database server, backups, webserver - which includes serving gentoo
portage and distfiles to other machines on the network (THTTPD is a
great minimal web server).

> What install docs did you follow?

Sheevaplug:

http://dev.gentoo.org/~armin76/arm/sheevaplug/install.xml#install

Pandaboard:

http://dev.gentoo.org/~armin76/arm/pandaboard/install.xml

It's easy.

> Any suggestions on setting up ARM servers, cluster,
> and such are most welcome.

ARM servers aren't much different to other servers but you must realise
that these are low powered devices (the ones I run anyway) and aren't
really suited to large loads. They especially suit a small business or
home hobbyist environment. Even so, compiling Gentoo, especially on the
Panda is not a problem and doesn't take forever (except for gcc
updates 8-)).

I suppose you could cluster a number of these devices but I think it
would be more efficient to use a more powerful server running servers as
virtual machines.

-- 
Regards,
Gregory.

[gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[James]

Gregory Shearman <zekeyg <at> gmail.com> writes:


> Both servers are running Gentoo Stable... therefore current kernels (for
> their architecture). Both have external HDD attached via USB.

Hey Greg,

If you just "reply" to the thread, we can keep one continuous 
thread going in lieu of a new posting each time.

Let's just look at the Panda board. I have a first rev panda to
experiment with. 

So a HDD via USB 2.0? fast enough for a Postgrsql database?
A bit more on the HDD setup (hardware) would be keen. 
Did you ever try to run this on a straight USB stick and not
the performance difference?


> File systems: root filesystem is on an SDHC card (2nd partition). Other
> filesystems (except for the boot partition) are all on LVM. I have
> /usr/src, /usr/portage, /usr/portage/distfiles is a symlink to
> /var/www/localhost/gentoo/distfiles (another filesystem). I also have
> /var/tmp/portage on a separate filesystem and I also run a postgresql
> database server which also has its own partition on
> /var/lib/postgresql/≤version>. Both servers have the same setup as I'm
> currently in the process of replacing the sheevaplug with the panda.

Postgresql on a separate partition, nice idea. Do you aggresively
manage  the PG server or is it just a recreational (light duty)
usage?

>
> Grub? There's no such thing on ARM machines. The kernel or uImage looks
> for the first partition on the configured root device (SDHC on my
> systems) the first partition MUST be VFAT (unfortunately) and it
> contains the u-boot bootloader and the kernel (uImage). 

https://wiki.linaro.org/LEG/Engineering/Grub2

https://wiki.linaro.org/LEG/Engineering/Kernel/ACPI/AcpiOnArndaleUefi

> Kernels are built the same way as x86 kernels except you do 
> "make uImage" instead of "make bzImage".

You Compile the kernels on a x86 host or compile them directly on
the Arm chip?

Then you put new kernels on the SD and swap those out to test/use
newer kernels on the Arm systems?


> LVM? All the above filesystems, except the root partition and the boot
> partition are LVM volumes. Filesystems are mostly Ext4 (very
> conventional).

What, no ZFS.....? Wait till Alan heards about this.....
Grub2 on ARM will allow many new file systems, and that
is the key issue with robust Arm servers, right now, imho.

> > Typical usage?

> Print server, database server, backups, webserver - which 
> includes serving gentoo portage and distfiles to other machines 
> on the network (THTTPD is a great minimal web server).

> > Any suggestions on setting up ARM servers, cluster,
> > and such are most welcome.

> ARM servers aren't much different to other servers but you must realise
> that these are low powered devices (the ones I run anyway) and aren't
> really suited to large loads. They especially suit a small business or
> home hobbyist environment. Even so, compiling Gentoo, especially on the
> Panda is not a problem and doesn't take forever (except for gcc
> updates ).

What does your make.conf look like on the panda?

> I suppose you could cluster a number of these devices but I think it
> would be more efficient to use a more powerful server running servers as
> virtual machines.


No BTRFS or CEPH?  (just teasing, but seriously....)

http://armservers.com/tag/ceph/
http://www.inktank.com/calxeda/

I posted previously on some Arm (A15) based systems, you may want to
look at for your next arm server, recently. Many have SATA 3 interfaces.

If you look at the ARM installation (handbook) docs, it is need of a
re_vamping.   I'm certain that folks would appreciate your
participation in the modernization of the ARM handbook, via the
Gentoo wiki.   The Gentoo wiki is your (ARM) friend....

I'm very happy, you are sharing your (ARM) gentoo experiences herein.


James

[gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[walt]

On 09/29/2013 04:58 AM, Volker Armin Hemmann wrote:

> As much as I hate systemd

My Alzheimer's prevents me from remembering your reasons for hating systemd.
Would you *very* briefly refresh my memory, please?

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Volker Armin Hemmann]

Am 08.10.2013 02:03, schrieb walt:
> On 09/29/2013 04:58 AM, Volker Armin Hemmann wrote:
>
>> As much as I hate systemd
> My Alzheimer's prevents me from remembering your reasons for hating systemd.
> Would you *very* briefly refresh my memory, please?
>
>
>
simple: one tool to do one job. text output to pipe into other tools.
Small is better.

systemd violates all of them. Also: dishonesty.

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[William Hubbs]

[-- Attachment #1: Type: text/plain, Size: 965 bytes --]

On Tue, Oct 08, 2013 at 08:11:48PM +0200, Volker Armin Hemmann wrote:
> Am 08.10.2013 02:03, schrieb walt:
> > On 09/29/2013 04:58 AM, Volker Armin Hemmann wrote:
> >
> >> As much as I hate systemd
> > My Alzheimer's prevents me from remembering your reasons for hating systemd.
> > Would you *very* briefly refresh my memory, please?
> >
> >
> >
> simple: one tool to do one job. text output to pipe into other tools.
> Small is better.

I'm not a strong systemd hater or anything, but this is my concern
about the way it is designed as well; process 1 is way too complex.

There is some interest in s6 [1], which is now in ~arch on amd64 and
x86. It seems to be a pretty simple design.

We haven't written anything for it yet, but it may be able to be
integrated into OpenRC to provide service supervision, which is the main
feature systemd offers, in my opinion, which we do not have in our
current OpenRC setup.

William

[1] http://www.skarnet.org/software/s6

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

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[gottlieb]

On Sun, Sep 29 2013, tanstaafl@libertytrek.org wrote:

> On 2013-09-29 2:55 PM, William Hubbs <williamh@gentoo.org> wrote:
>> I am the OpenRC author/maintainer and a member of base-system. I can
>> tell you that we are not discussing forcing systemd on everyone in
>> Gentoo Linux as a default init system. I can also tell you that I am not
>> aware of the Gentoo systemd team discussing this. Even if they were, a
>> distro-wide change like this would have to be brought before the
>> Council.
>
> Ok, good enough for me until other evidence comes along to cast doubt
> as to the truthfulness or sincerity of your statement.
>
> Thanks William...
>
> Now to try to get up enough nerve to attempt to merge my /usr
> (currently on LVM partition) into my / (does have enough room, and
> will leave me with a 19GB / partition with about 5GB free).
>
> Anyone see a problem with that (only 5GB free on my / after the /usr merge)?

I understand the need to get up nerve.  That was the hardest part for
me, and took by far, the most time.  I did *not* have room in / for /usr
but *did* have an online external disk on the machine with lots of room
(Alan's "what I should have done" scheme).  I could afford downtime so I
did everything booted from an installation CD so that nothing would
change.

1.  Booted minimal installation CD
2.  Copied my 5 lvs (/usr, /opt, /var, /tmp, /local) 
    and my / to the external disk and called them old-root, old-usr,
    old-opt, old-var, old-tmp, old-local.
3.  Repartitioned the internal disk to make root bigger.
4.  Created the vg and pv (I have just one of each).
5.  Created the 5 filesystems (root, /opt, /var, /tmp, /local), with the
    last 4 on LVM
6.  Copied old-root to / and old-usr to /usr
7.  Mounted the 4 lvs and copied old-opt to /opt, old-var to /var, ...

Reboot

It worked.

Notes.

1.  I had grub in the MBR so that didn't change
2.  The root fs remained the same partition number (/dev/sda3),
    so didn't have to change grub.
3.  In fact /dev/sda3 maintained the same starting location in the new
    partitioning scheme, but I don't think that was relevant.

allan

[gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[walt]

On 10/08/2013 09:16 PM, William Hubbs wrote:

> to provide service supervision, which is the main
> feature systemd offers

By supervision do you mean restarting a service after it crashes, for example?

Or something else completely?

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[William Hubbs]

[-- Attachment #1: Type: text/plain, Size: 373 bytes --]

On Wed, Oct 09, 2013 at 05:24:39PM -0700, walt wrote:
> On 10/08/2013 09:16 PM, William Hubbs wrote:
> 
> > to provide service supervision, which is the main
> > feature systemd offers
> 
> By supervision do you mean restarting a service after it crashes, for example?

Right. This is one of the more significant features that OpenRC doesn't
have yet.

William

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

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Volker Armin Hemmann]

Am 10.10.2013 16:46, schrieb William Hubbs:
> On Wed, Oct 09, 2013 at 05:24:39PM -0700, walt wrote:
>> On 10/08/2013 09:16 PM, William Hubbs wrote:
>>
>>> to provide service supervision, which is the main
>>> feature systemd offers
>> By supervision do you mean restarting a service after it crashes, for example?
> Right. This is one of the more significant features that OpenRC doesn't
> have yet.
>
> William

why?

if something like sshd crashes, you either have a hardware problem or
sshd is buggy. Either way, better not be pampered over with a silent
service restart.

The rest is so visible (or audible - like fancontrol) that you know that
there is a problem.

[gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Nicolas Sebrecht]

The 10/10/13, Volker Armin Hemmann wrote:

> if something like sshd crashes, you either have a hardware problem or
> sshd is buggy. Either way, better not be pampered over with a silent
> service restart.

So, restarting a service should not be silent (I think it isn't) and
might need better alerts. Oh, don't the admin have the tools for this
already (sendmail, motd, snmp, whatever)?

I'm not pretending the current situation is perfect but if admins are
tired to configure alerts on their own, it should not be that hard to
improve and factorize efforts (at Gentoo at least, if not upstream).

-- 
Nicolas Sebrecht

[gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Steven J. Long]

On Sat, Sep 28, 2013 at 09:17:02PM +0100, Neil Bothwick wrote:
> On Sat, 28 Sep 2013 19:04:41 +0000, Alan Mackenzie wrote:
> 
> > > I suppose that what I am about to say isn't really relevant, but it is
> > > unfortunate over the past year that people blamed udev specifically
> > > for this. It is true that it does things that don't work if /usr isn't
> > > mounted, but eudev does as well, since it is basically the same code.
> > 
> > Who else is there to blame?  We are continually being told that a
> > separate /usr is "broken", as though this were some unfortunate act of
> > <insert your deity here>, much like an earthquake.  This gets
> > patronising really quickly.  (Please note, I'm NOT blaming you here.  I
> > appreciate that you're as much victim as Dale or me or anyone else
> > round here.)
> 
> It's evolution. Linux has for years been moving in this direction, now it
> has reached the point where the Gentoo devs can no longer devote the
> increasing time needed to support what has now become an dge case.

Yeah and that's just vague crap without content ;)

> > No, this breaking of separate /usr was done by some specific project,
> > some specific person, even, in a supreme display of incompetence,
> > malice, or arrogance.  How come this project and this person have
> > managed to maintain such a low profile?  There seems to have been some
> > sort of conspiracy to do this breakage in secret, each member of the
> > coven pushing the plot until the damage was irrevocable.  Who was it?
> 
> So which was it, one specific person or a coven of conspirators? This is
> open source, secret conspiracies don't really work well. If this really
> was such a bad move, do you really think the likes of Greg K-H would not
> have stepped in? Or is he a conspirator too?

No he's just a bit naive: he wants to believe the best of people and did
not realise quite how sneaky Poettering is. No doubt he still doesn't. But
I'm sure he never foresaw some of their shenanighans, such as claiming their
newly inserted breakage was the fault of device-drivers and everyone should
switch to their funky new way of loading modules. No-one seemed to think
what Torvalds said was incorrect, even if they disagreed with his tone.

And yet that's exactly the same crap they pull in user-space, only they seem
to think the kernel mentality of "userspace is crazy" is a howto methodology.

-- 
#friendly-coders -- We're friendly, but we're not /that/ friendly ;-)

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 1642 bytes --]

On Fri, 11 Oct 2013 09:36:02 +0100, Steven J. Long wrote:

> > It's evolution. Linux has for years been moving in this direction,
> > now it has reached the point where the Gentoo devs can no longer
> > devote the increasing time needed to support what has now become an
> > dge case.  
> 
> Yeah and that's just vague crap without content ;)

I bow to your superior expertise in that field :)

> > So which was it, one specific person or a coven of conspirators? This
> > is open source, secret conspiracies don't really work well. If this
> > really was such a bad move, do you really think the likes of Greg K-H
> > would not have stepped in? Or is he a conspirator too?  
> 
> No he's just a bit naive: he wants to believe the best of people and did
> not realise quite how sneaky Poettering is. No doubt he still doesn't.
> But I'm sure he never foresaw some of their shenanighans, such as
> claiming their newly inserted breakage was the fault of device-drivers
> and everyone should switch to their funky new way of loading modules.
> No-one seemed to think what Torvalds said was incorrect, even if they
> disagreed with his tone.

I don't understand why people keep banging on about Poettering in this,
previously finished, thread. The announcement was made by the OpenRC
maintainer and applies equally to those running eudev as udev. That is,
systems free of that individual's influence. Whatever anyone's opinion of
the way he is taking things, and for the record I don't like systemd,
this is a situation that arose without his help.


-- 
Neil Bothwick

Multitasking: Reading in the bathroom.

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

[gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[walt]

On 10/11/2013 01:42 AM, Neil Bothwick wrote:

> I don't like systemd,

Sorry if my memory is failing (it surely is) but I don't recall any
explanation from you describing your dissatisfaction with systemd.

The three happiest months of my life were spent as a student in London
in the summer of 1974, where I frequently heard the phrase "I should have
thought that you...".

Only much later did I discover that such a benign phrase conveys the most
severe form of British disapproval :(

With belated apologies to my many kind Brit friends from 1974, I ask you
to tell us WTF you dislike systemd, and use language that us Yanks can
fscking unnerstand, got it, punk?

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Daniel Campbell]

On 10/11/2013 09:21 PM, walt wrote:
> On 10/11/2013 01:42 AM, Neil Bothwick wrote:
> 
>> I don't like systemd,
> 
> Sorry if my memory is failing (it surely is) but I don't recall any
> explanation from you describing your dissatisfaction with systemd.
> 
> The three happiest months of my life were spent as a student in London
> in the summer of 1974, where I frequently heard the phrase "I should have
> thought that you...".
> 
> Only much later did I discover that such a benign phrase conveys the most
> severe form of British disapproval :(
> 
> With belated apologies to my many kind Brit friends from 1974, I ask you
> to tell us WTF you dislike systemd, and use language that us Yanks can
> fscking unnerstand, got it, punk?
> 
> 
What do his personal opinions regarding systemd have to do with separate
/ and /usr? It's just another one of many, many applications that
migrated to /usr and added more inertia to de facto practice.

Re: [gentoo-user] Re: separate / and /usr to require initramfs 2013-11-01

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 1455 bytes --]

On Fri, 11 Oct 2013 19:21:05 -0700, walt wrote:

> On 10/11/2013 01:42 AM, Neil Bothwick wrote:
> 
> > I don't like systemd,
> 
> Sorry if my memory is failing (it surely is) but I don't recall any
> explanation from you describing your dissatisfaction with systemd.

It was never germane to the conversation. I only mentioned it here to
make it clear that I am not a systemd or Poettering apologist.

I don't like the idea of such a complex and pervasive init process. Do
one thing and do it well is the long-standing Unix mantra, and it's been
long-standing for good reason. This is particularly applicable to the
most critical process on the system, process 1.

I'm also uncomfortable with the close ties between systemd and GNOME, not
that  have anything against the GNOME people but init should be
independently controlled. Red Hat contribute more to the kernel than
anyone else (12.5% IIRC) but they don't control its development.

I have tried systemd on a minimal VM and it did boot very quickly, but
that's not a real concern for me. The only system I reboot with any
regularity is my laptop, and that boots equally quickly because it has an
SSD.

> With belated apologies to my many kind Brit friends from 1974, I ask you
> to tell us WTF you dislike systemd, and use language that us Yanks can
> fscking unnerstand, got it, punk?

OMG IT'S NOT AWESOME!


-- 
Neil Bothwick

New sig wanted good price paid.

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