install Fedora systemd-nspawn container on btrfs

install Fedora systemd-nspawn container on btrfs

[arnaud gaboury]

Not sure what I did wrong, but I can't install/boot my nspawn container.
Here is my setup:

Archlinux box- updated

1- created 3 btrfs subvol on /dev/sdb1 (SSD). The goal is to manage
snapshots easily.
no nested subvol.
------------------------------------------
# btrfs subvolume list .
ID 266 gen 39 top level 5 path rootvol
ID 268 gen 41 top level 5 path var
ID 269 gen 42 top level 5 path etc
# btrfs filesystem show
Label: 'poppy-root'  uuid: ef1b44cd-e7b0-4166-b933-e7d4d20a1171
    Total devices 1 FS bytes used 64.00KiB
    devid    1 size 80.00GiB used 12.00MiB path /dev/sdb1
--------------------------------------------------

2 - mount btrfs subvol
-------------------------------------------------------------------
# mount -t btrfs -o subvol=rootvol /dev/sdb1 /var/lib/machines/enl
# mkdir /var/lib/machines/enl/var
# mkdir /var/lib/machines/enl/etc
# mount -t btrfs -o subvol=etc /dev/sdb1 /var/lib/machines/enl/etc
# mount -t btrfs -o subvol=var /dev/sdb1 /var/lib/machines/enl/var
--------------------------------------------------------------------

3- install fedora minimal and boot it
-------------------------------------------------
#  machinectl pull-raw --verify=no
http://ftp.halifax.rwth-aachen.de/fedora/linux/releases/22/Cloud/Images/x86_64/Fedora-Cloud-Base-22_Beta-20150415-x86_64.raw.xz
$ tar ....
# systemd-nspawn -M Fedora-Cloud-Base-22_Beta-20150415.x86_64.raw
Spawning container Fedora-Cloud-Base-22_Beta-20150415.x86_64.raw on
/var/lib/machines/Fedora-Cloud-Base-22_Beta-20150415.x86_64.raw.
Press ^] three times within 1s to kill container.
[root@Fedora-Cloud-Base-22_Beta-20150415 ~]#
------------------------------------------------------------

4- install Fedora on /var/lib/machines/enl
--------------------------------------------------------------
[root@Fedora-Cloud-Base-22_Beta-20150415 ~]# dnf -y --releasever=22
--nogpg --installroot=/var/lib/machines/MyContainer --disablerepo='*'
--enablerepo=fedora install systemd passwd dnf fedora-release-server
vim-minimal
..........................
INSTALL
...........................
Complete!
----------------------------------------------------------

5- Now, nothing !!
----------------------------------------------------
% ls -a /var/lib/machines/enl
./  ../  etc/  var/

# systemd-nspawn -bD /var/lib/machines/enl
[sudo] password for root:
Directory /var/lib/machines/enl doesn't look like an OS root directory
(os-release file is missing). Refusing.
-------------------------------------------------------

Not sure if my issue comes from a wrong Btrfs setup, or if I missed
something when installing from the cloud minimal Fedora.

Thank you for hints







-- 

google.com/+arnaudgabourygabx

Re: [systemd-devel] install Fedora systemd-nspawn container on btrfs

[Lennart Poettering]

On Thu, 23.04.15 13:45, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:

> Not sure what I did wrong, but I can't install/boot my nspawn container.
> Here is my setup:
> 
> Archlinux box- updated
> 
> 1- created 3 btrfs subvol on /dev/sdb1 (SSD). The goal is to manage
> snapshots easily.
> no nested subvol.
> ------------------------------------------
> # btrfs subvolume list .
> ID 266 gen 39 top level 5 path rootvol
> ID 268 gen 41 top level 5 path var
> ID 269 gen 42 top level 5 path etc
> # btrfs filesystem show
> Label: 'poppy-root'  uuid: ef1b44cd-e7b0-4166-b933-e7d4d20a1171
>     Total devices 1 FS bytes used 64.00KiB
>     devid    1 size 80.00GiB used 12.00MiB path /dev/sdb1
> --------------------------------------------------
> 
> 2 - mount btrfs subvol
> -------------------------------------------------------------------
> # mount -t btrfs -o subvol=rootvol /dev/sdb1 /var/lib/machines/enl
> # mkdir /var/lib/machines/enl/var
> # mkdir /var/lib/machines/enl/etc
> # mount -t btrfs -o subvol=etc /dev/sdb1 /var/lib/machines/enl/etc
> # mount -t btrfs -o subvol=var /dev/sdb1 /var/lib/machines/enl/var
> --------------------------------------------------------------------

THis isn't really how one would normally use subvolumes. No need to
mount each subvolume explicitly, they are just special directories...

> 
> 3- install fedora minimal and boot it
> -------------------------------------------------
> #  machinectl pull-raw --verify=no
> http://ftp.halifax.rwth-aachen.de/fedora/linux/releases/22/Cloud/Images/x86_64/Fedora-Cloud-Base-22_Beta-20150415-x86_64.raw.xz
> $ tar ....
> # systemd-nspawn -M Fedora-Cloud-Base-22_Beta-20150415.x86_64.raw
> Spawning container Fedora-Cloud-Base-22_Beta-20150415.x86_64.raw on
> /var/lib/machines/Fedora-Cloud-Base-22_Beta-20150415.x86_64.raw.
> Press ^] three times within 1s to kill container.
> [root@Fedora-Cloud-Base-22_Beta-20150415 ~]#
> ------------------------------------------------------------
> 
> 4- install Fedora on /var/lib/machines/enl
> --------------------------------------------------------------
> [root@Fedora-Cloud-Base-22_Beta-20150415 ~]# dnf -y --releasever=22
> --nogpg --installroot=/var/lib/machines/MyContainer --disablerepo='*'
> --enablerepo=fedora install systemd passwd dnf fedora-release-server
> vim-minimal
> ..........................
> INSTALL
> ...........................
> Complete!
> ----------------------------------------------------------


Hmm? With this command you installed another fedora inside the raw
fedora image you downloaded. You now have three linuxes, installed
within each other...

Pick one:

a) download the raw image and use that, but it will be a loopback file
with its own file system inside

or:

b) do the dnf/yum install root thing, and install it into a directory
tree.

Do either of those in the host, not in the container.

Lennart

-- 
Lennart Poettering, Red Hat

Re: [systemd-devel] install Fedora systemd-nspawn container on btrfs

[Andrei Borzenkov]

On Thu, Apr 23, 2015 at 2:50 PM, Lennart Poettering
<lennart@poettering.net> wrote:
> On Thu, 23.04.15 13:45, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:
>
>> Not sure what I did wrong, but I can't install/boot my nspawn container.
>> Here is my setup:
>>
>> Archlinux box- updated
>>
>> 1- created 3 btrfs subvol on /dev/sdb1 (SSD). The goal is to manage
>> snapshots easily.
>> no nested subvol.
>> ------------------------------------------
>> # btrfs subvolume list .
>> ID 266 gen 39 top level 5 path rootvol
>> ID 268 gen 41 top level 5 path var
>> ID 269 gen 42 top level 5 path etc
>> # btrfs filesystem show
>> Label: 'poppy-root'  uuid: ef1b44cd-e7b0-4166-b933-e7d4d20a1171
>>     Total devices 1 FS bytes used 64.00KiB
>>     devid    1 size 80.00GiB used 12.00MiB path /dev/sdb1
>> --------------------------------------------------
>>
>> 2 - mount btrfs subvol
>> -------------------------------------------------------------------
>> # mount -t btrfs -o subvol=rootvol /dev/sdb1 /var/lib/machines/enl
>> # mkdir /var/lib/machines/enl/var
>> # mkdir /var/lib/machines/enl/etc
>> # mount -t btrfs -o subvol=etc /dev/sdb1 /var/lib/machines/enl/etc
>> # mount -t btrfs -o subvol=var /dev/sdb1 /var/lib/machines/enl/var
>> --------------------------------------------------------------------
>
> THis isn't really how one would normally use subvolumes. No need to
> mount each subvolume explicitly, they are just special directories...
>

As long as you never clone parent volume (but why use btrfs then?) As
soon as you create clone or snapshot of parent volume, all childs will
be out of place in it unless you explicitly mount them in correct
place in hierarchy.

Re: [systemd-devel] install Fedora systemd-nspawn container on btrfs

[Lennart Poettering]

On Thu, 23.04.15 14:57, Andrei Borzenkov (arvidjaar@gmail.com) wrote:

> On Thu, Apr 23, 2015 at 2:50 PM, Lennart Poettering
> <lennart@poettering.net> wrote:
> > On Thu, 23.04.15 13:45, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:
> >
> >> Not sure what I did wrong, but I can't install/boot my nspawn container.
> >> Here is my setup:
> >>
> >> Archlinux box- updated
> >>
> >> 1- created 3 btrfs subvol on /dev/sdb1 (SSD). The goal is to manage
> >> snapshots easily.
> >> no nested subvol.
> >> ------------------------------------------
> >> # btrfs subvolume list .
> >> ID 266 gen 39 top level 5 path rootvol
> >> ID 268 gen 41 top level 5 path var
> >> ID 269 gen 42 top level 5 path etc
> >> # btrfs filesystem show
> >> Label: 'poppy-root'  uuid: ef1b44cd-e7b0-4166-b933-e7d4d20a1171
> >>     Total devices 1 FS bytes used 64.00KiB
> >>     devid    1 size 80.00GiB used 12.00MiB path /dev/sdb1
> >> --------------------------------------------------
> >>
> >> 2 - mount btrfs subvol
> >> -------------------------------------------------------------------
> >> # mount -t btrfs -o subvol=rootvol /dev/sdb1 /var/lib/machines/enl
> >> # mkdir /var/lib/machines/enl/var
> >> # mkdir /var/lib/machines/enl/etc
> >> # mount -t btrfs -o subvol=etc /dev/sdb1 /var/lib/machines/enl/etc
> >> # mount -t btrfs -o subvol=var /dev/sdb1 /var/lib/machines/enl/var
> >> --------------------------------------------------------------------
> >
> > THis isn't really how one would normally use subvolumes. No need to
> > mount each subvolume explicitly, they are just special directories...
> 
> As long as you never clone parent volume (but why use btrfs then?) As
> soon as you create clone or snapshot of parent volume, all childs will
> be out of place in it unless you explicitly mount them in correct
> place in hierarchy.

Hmm? not following. The "btrfs" tool surely doesn't do recursive
snapshots currently. But it doesn't reinstate mount points either (or
even makes them persistent), hence I really don't get what you are
saying.

(note that machined's clone command in git *does* recursive snapshots)

Lennart

-- 
Lennart Poettering, Red Hat

Re: [systemd-devel] install Fedora systemd-nspawn container on btrfs

[arnaud gaboury]

On Thu, Apr 23, 2015 at 2:00 PM, Lennart Poettering
<lennart@poettering.net> wrote:
> On Thu, 23.04.15 14:57, Andrei Borzenkov (arvidjaar@gmail.com) wrote:
>
>> On Thu, Apr 23, 2015 at 2:50 PM, Lennart Poettering
>> <lennart@poettering.net> wrote:
>> > On Thu, 23.04.15 13:45, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:
>> >
>> >> Not sure what I did wrong, but I can't install/boot my nspawn container.
>> >> Here is my setup:
>> >>
>> >> Archlinux box- updated
>> >>
>> >> 1- created 3 btrfs subvol on /dev/sdb1 (SSD). The goal is to manage
>> >> snapshots easily.
>> >> no nested subvol.
>> >> ------------------------------------------
>> >> # btrfs subvolume list .
>> >> ID 266 gen 39 top level 5 path rootvol
>> >> ID 268 gen 41 top level 5 path var
>> >> ID 269 gen 42 top level 5 path etc
>> >> # btrfs filesystem show
>> >> Label: 'poppy-root'  uuid: ef1b44cd-e7b0-4166-b933-e7d4d20a1171
>> >>     Total devices 1 FS bytes used 64.00KiB
>> >>     devid    1 size 80.00GiB used 12.00MiB path /dev/sdb1
>> >> --------------------------------------------------
>> >>
>> >> 2 - mount btrfs subvol
>> >> -------------------------------------------------------------------
>> >> # mount -t btrfs -o subvol=rootvol /dev/sdb1 /var/lib/machines/enl
>> >> # mkdir /var/lib/machines/enl/var
>> >> # mkdir /var/lib/machines/enl/etc
>> >> # mount -t btrfs -o subvol=etc /dev/sdb1 /var/lib/machines/enl/etc
>> >> # mount -t btrfs -o subvol=var /dev/sdb1 /var/lib/machines/enl/var
>> >> --------------------------------------------------------------------
>> >
>> > THis isn't really how one would normally use subvolumes. No need to
>> > mount each subvolume explicitly, they are just special directories...
>>
>> As long as you never clone parent volume (but why use btrfs then?) As
>> soon as you create clone or snapshot of parent volume, all childs will
>> be out of place in it unless you explicitly mount them in correct
>> place in hierarchy.
>
> Hmm? not following. The "btrfs" tool surely doesn't do recursive
> snapshots currently. But it doesn't reinstate mount points either (or
> even makes them persistent), hence I really don't get what you are
> saying.
>
> (note that machined's clone command in git *does* recursive snapshots)

I am not a Btrfs expert, but from what I understood, and after some
posts on Btrfs ML[0], creating and mounting these subvol was the right
way to easily manage snapshots.

[0]http://www.mail-archive.com/linux-btrfs@vger.kernel.org/msg42856.html
>
> Lennart
>
> --
> Lennart Poettering, Red Hat



-- 

google.com/+arnaudgabourygabx

Re: [systemd-devel] install Fedora systemd-nspawn container on btrfs

[arnaud gaboury]

On Thu, Apr 23, 2015 at 1:50 PM, Lennart Poettering
<lennart@poettering.net> wrote:
> On Thu, 23.04.15 13:45, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:
>
>> Not sure what I did wrong, but I can't install/boot my nspawn container.
>> Here is my setup:
>>
>> Archlinux box- updated
>>
>> 1- created 3 btrfs subvol on /dev/sdb1 (SSD). The goal is to manage
>> snapshots easily.
>> no nested subvol.
>> ------------------------------------------
>> # btrfs subvolume list .
>> ID 266 gen 39 top level 5 path rootvol
>> ID 268 gen 41 top level 5 path var
>> ID 269 gen 42 top level 5 path etc
>> # btrfs filesystem show
>> Label: 'poppy-root'  uuid: ef1b44cd-e7b0-4166-b933-e7d4d20a1171
>>     Total devices 1 FS bytes used 64.00KiB
>>     devid    1 size 80.00GiB used 12.00MiB path /dev/sdb1
>> --------------------------------------------------
>>
>> 2 - mount btrfs subvol
>> -------------------------------------------------------------------
>> # mount -t btrfs -o subvol=rootvol /dev/sdb1 /var/lib/machines/enl
>> # mkdir /var/lib/machines/enl/var
>> # mkdir /var/lib/machines/enl/etc
>> # mount -t btrfs -o subvol=etc /dev/sdb1 /var/lib/machines/enl/etc
>> # mount -t btrfs -o subvol=var /dev/sdb1 /var/lib/machines/enl/var
>> --------------------------------------------------------------------
>
> THis isn't really how one would normally use subvolumes. No need to
> mount each subvolume explicitly, they are just special directories...
>
>>
>> 3- install fedora minimal and boot it
>> -------------------------------------------------
>> #  machinectl pull-raw --verify=no
>> http://ftp.halifax.rwth-aachen.de/fedora/linux/releases/22/Cloud/Images/x86_64/Fedora-Cloud-Base-22_Beta-20150415-x86_64.raw.xz
>> $ tar ....
>> # systemd-nspawn -M Fedora-Cloud-Base-22_Beta-20150415.x86_64.raw
>> Spawning container Fedora-Cloud-Base-22_Beta-20150415.x86_64.raw on
>> /var/lib/machines/Fedora-Cloud-Base-22_Beta-20150415.x86_64.raw.
>> Press ^] three times within 1s to kill container.
>> [root@Fedora-Cloud-Base-22_Beta-20150415 ~]#
>> ------------------------------------------------------------
>>
>> 4- install Fedora on /var/lib/machines/enl
>> --------------------------------------------------------------
>> [root@Fedora-Cloud-Base-22_Beta-20150415 ~]# dnf -y --releasever=22
>> --nogpg --installroot=/var/lib/machines/MyContainer --disablerepo='*'
>> --enablerepo=fedora install systemd passwd dnf fedora-release-server
>> vim-minimal
>> ..........................
>> INSTALL
>> ...........................
>> Complete!
>> ----------------------------------------------------------
>
>
> Hmm? With this command you installed another fedora inside the raw
> fedora image you downloaded. You now have three linuxes, installed
> within each other...

I suspected such kind of thing indeed. In the container,
/var/lib/machines is in fact inside the container and not on my host!
Stupid.
>
> Pick one:
>
> a) download the raw image and use that, but it will be a loopback file
> with its own file system inside
>
> or:
>
> b) do the dnf/yum install root thing, and install it into a directory
> tree.

I installed yum package on Arch but couldn't manage to do the install.

# yum -y --releasever=22 --nogpg --installroot=/var/lib/machines/enl
--disablerepo='*' --enablerepo=fedora install systemd passwd dnf
fedora-release-server
Error getting repository data for fedora, repository not found
# yum repolist all
repolist: 0

In fact, /etc/yum/repos.d is empty, so I am not surprised.



>
> Do either of those in the host, not in the container.
>
> Lennart
>
> --
> Lennart Poettering, Red Hat



-- 

google.com/+arnaudgabourygabx

Re: [systemd-devel] install Fedora systemd-nspawn container on btrfs

[Lennart Poettering]

On Thu, 23.04.15 14:18, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:

> > Pick one:
> >
> > a) download the raw image and use that, but it will be a loopback file
> > with its own file system inside
> >
> > or:
> >
> > b) do the dnf/yum install root thing, and install it into a directory
> > tree.
> 
> I installed yum package on Arch but couldn't manage to do the install.
> 
> # yum -y --releasever=22 --nogpg --installroot=/var/lib/machines/enl
> --disablerepo='*' --enablerepo=fedora install systemd passwd dnf
> fedora-release-server
> Error getting repository data for fedora, repository not found
> # yum repolist all
> repolist: 0
> 
> In fact, /etc/yum/repos.d is empty, so I am not surprised.

Of course, you could use the .raw image, mount the extern btrfs volume
into it via nspawn's --bind= switch, then use yum inside of that
container to install into the btrfs volume. Then get rid of the .raw
image again, and you still have the btrfs volume that should be
bootable.

A bit complex, but you almost were there already... ;-)

Lennart

-- 
Lennart Poettering, Red Hat

Re: [systemd-devel] install Fedora systemd-nspawn container on btrfs

[arnaud gaboury]

On Thu, Apr 23, 2015 at 4:47 PM, Lennart Poettering
<lennart@poettering.net> wrote:
> On Thu, 23.04.15 14:18, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:
>
>> > Pick one:
>> >
>> > a) download the raw image and use that, but it will be a loopback file
>> > with its own file system inside
>> >
>> > or:
>> >
>> > b) do the dnf/yum install root thing, and install it into a directory
>> > tree.
>>
>> I installed yum package on Arch but couldn't manage to do the install.
>>
>> # yum -y --releasever=22 --nogpg --installroot=/var/lib/machines/enl
>> --disablerepo='*' --enablerepo=fedora install systemd passwd dnf
>> fedora-release-server
>> Error getting repository data for fedora, repository not found
>> # yum repolist all
>> repolist: 0
>>
>> In fact, /etc/yum/repos.d is empty, so I am not surprised.
>
> Of course, you could use the .raw image, mount the extern btrfs volume
> into it via nspawn's --bind= switch, then use yum inside of that
> container to install into the btrfs volume. Then get rid of the .raw
> image again, and you still have the btrfs volume that should be
> bootable.
>
> A bit complex, but you almost were there already... ;-)

Wunderbach

# systemd-nspawn -M Fedora-Cloud-Base-22_Beta-20150415.x86_64.raw --
bind=/var/lib/machines/enl:/mnt
[root@Fedora-Cloud-Base-22_Beta-20150415 ~]#dnf -y --releasever=22
--nogpg --installroot=/mnt
 --disablerepo='*' --enablerepo=fedora install systemd passwd dnf
................
Complete!

-------------------------------------------------------------------------
$ ls /var/lib/machines/enl
boot/  etc/   media/  opt/   root/  srv/  tmp/  var/  lib@    sbin@
dev/   home/  mnt/    proc/  run/   sys/  usr/  bin@  lib64@
-------------------------------------------------------------------------------------

But now booting but hanging:

#systemd-nspawn -bD /var/lib/machines/enl
Spawning container enl on /var/lib/machines/enl.
Press ^] three times within 1s to kill container.
systemd 219 running in system mode. (+PAM +AUDIT +SELINUX +IMA
-APPARMOR +SMACK +SYSVINIT +UTMP +LIBCRYPTSETUP +GCRYPT +GNUTLS +ACL
+XZ -LZ4 +SECCOMP +BLKID +ELFUTILS +KMOD +IDN)
Detected virtualization 'systemd-nspawn'.
Detected architecture 'x86-64'.
Running with unpopulated /etc.

Welcome to Fedora 22 (Twenty Two)!

Initializing machine ID from random generator.
Populated /etc with preset unit settings.
Unit etc.mount is bound to inactive unit dev-sdb1.device. Stopping, too.
Unit var.mount is bound to inactive unit dev-sdb1.device. Stopping, too.
Cannot add dependency job for unit display-manager.service, ignoring:
Unit display-manager.service failed to load: No such file or
directory.
Startup finished in 51ms.

------------------------------------------------------------------------

Maybe my btrfs story ?
> Unit etc.mount is bound to inactive unit dev-sdb1.device. Stopping, too.
> Unit var.mount is bound to inactive unit dev-sdb1.device. Stopping, too.
Will investigate.

--bind was certainly the most easy trick once the raw image is downloaded.

>
> Lennart
>
> --
> Lennart Poettering, Red Hat



-- 

google.com/+arnaudgabourygabx

Re: [systemd-devel] install Fedora systemd-nspawn container on btrfs

[Lennart Poettering]

On Thu, 23.04.15 19:00, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:

> On Thu, Apr 23, 2015 at 4:47 PM, Lennart Poettering
> <lennart@poettering.net> wrote:
> > On Thu, 23.04.15 14:18, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:
> >
> >> > Pick one:
> >> >
> >> > a) download the raw image and use that, but it will be a loopback file
> >> > with its own file system inside
> >> >
> >> > or:
> >> >
> >> > b) do the dnf/yum install root thing, and install it into a directory
> >> > tree.
> >>
> >> I installed yum package on Arch but couldn't manage to do the install.
> >>
> >> # yum -y --releasever=22 --nogpg --installroot=/var/lib/machines/enl
> >> --disablerepo='*' --enablerepo=fedora install systemd passwd dnf
> >> fedora-release-server
> >> Error getting repository data for fedora, repository not found
> >> # yum repolist all
> >> repolist: 0
> >>
> >> In fact, /etc/yum/repos.d is empty, so I am not surprised.
> >
> > Of course, you could use the .raw image, mount the extern btrfs volume
> > into it via nspawn's --bind= switch, then use yum inside of that
> > container to install into the btrfs volume. Then get rid of the .raw
> > image again, and you still have the btrfs volume that should be
> > bootable.
> >
> > A bit complex, but you almost were there already... ;-)
> 
> Wunderbach
> 
> # systemd-nspawn -M Fedora-Cloud-Base-22_Beta-20150415.x86_64.raw --
> bind=/var/lib/machines/enl:/mnt
> [root@Fedora-Cloud-Base-22_Beta-20150415 ~]#dnf -y --releasever=22
> --nogpg --installroot=/mnt
>  --disablerepo='*' --enablerepo=fedora install systemd passwd dnf
> ................
> Complete!
> 
> -------------------------------------------------------------------------
> $ ls /var/lib/machines/enl
> boot/  etc/   media/  opt/   root/  srv/  tmp/  var/  lib@    sbin@
> dev/   home/  mnt/    proc/  run/   sys/  usr/  bin@  lib64@
> -------------------------------------------------------------------------------------
> 
> But now booting but hanging:
> 
> #systemd-nspawn -bD /var/lib/machines/enl
> Spawning container enl on /var/lib/machines/enl.
> Press ^] three times within 1s to kill container.
> systemd 219 running in system mode. (+PAM +AUDIT +SELINUX +IMA
> -APPARMOR +SMACK +SYSVINIT +UTMP +LIBCRYPTSETUP +GCRYPT +GNUTLS +ACL
> +XZ -LZ4 +SECCOMP +BLKID +ELFUTILS +KMOD +IDN)
> Detected virtualization 'systemd-nspawn'.
> Detected architecture 'x86-64'.
> Running with unpopulated /etc.
> 
> Welcome to Fedora 22 (Twenty Two)!
> 
> Initializing machine ID from random generator.
> Populated /etc with preset unit settings.
> Unit etc.mount is bound to inactive unit dev-sdb1.device. Stopping, too.
> Unit var.mount is bound to inactive unit dev-sdb1.device. Stopping, too.
> Cannot add dependency job for unit display-manager.service, ignoring:
> Unit display-manager.service failed to load: No such file or
> directory.
> Startup finished in 51ms.
> 
> ------------------------------------------------------------------------
> 
> Maybe my btrfs story ?
> > Unit etc.mount is bound to inactive unit dev-sdb1.device. Stopping, too.
> > Unit var.mount is bound to inactive unit dev-sdb1.device. Stopping, too.
> Will investigate.
> 
> --bind was certainly the most easy trick once the raw image is downloaded.

Hmm, my guess is that you somehow lost the /etc and /var directories
half way, probably because of the weird mounting your are
doing. --bind= should normally be recursive, but maybe this didn't work?

Lennart

-- 
Lennart Poettering, Red Hat

Re: [systemd-devel] install Fedora systemd-nspawn container on btrfs

[arnaud gaboury]

On Thu, Apr 23, 2015 at 7:12 PM, Lennart Poettering
<lennart@poettering.net> wrote:
> On Thu, 23.04.15 19:00, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:
>
>> On Thu, Apr 23, 2015 at 4:47 PM, Lennart Poettering
>> <lennart@poettering.net> wrote:
>> > On Thu, 23.04.15 14:18, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:
>> >
>> >> > Pick one:
>> >> >
>> >> > a) download the raw image and use that, but it will be a loopback file
>> >> > with its own file system inside
>> >> >
>> >> > or:
>> >> >
>> >> > b) do the dnf/yum install root thing, and install it into a directory
>> >> > tree.
>> >>
>> >> I installed yum package on Arch but couldn't manage to do the install.
>> >>
>> >> # yum -y --releasever=22 --nogpg --installroot=/var/lib/machines/enl
>> >> --disablerepo='*' --enablerepo=fedora install systemd passwd dnf
>> >> fedora-release-server
>> >> Error getting repository data for fedora, repository not found
>> >> # yum repolist all
>> >> repolist: 0
>> >>
>> >> In fact, /etc/yum/repos.d is empty, so I am not surprised.
>> >
>> > Of course, you could use the .raw image, mount the extern btrfs volume
>> > into it via nspawn's --bind= switch, then use yum inside of that
>> > container to install into the btrfs volume. Then get rid of the .raw
>> > image again, and you still have the btrfs volume that should be
>> > bootable.
>> >
>> > A bit complex, but you almost were there already... ;-)
>>
>> Wunderbach
>>
>> # systemd-nspawn -M Fedora-Cloud-Base-22_Beta-20150415.x86_64.raw --
>> bind=/var/lib/machines/enl:/mnt
>> [root@Fedora-Cloud-Base-22_Beta-20150415 ~]#dnf -y --releasever=22
>> --nogpg --installroot=/mnt
>>  --disablerepo='*' --enablerepo=fedora install systemd passwd dnf
>> ................
>> Complete!
>>
>> -------------------------------------------------------------------------
>> $ ls /var/lib/machines/enl
>> boot/  etc/   media/  opt/   root/  srv/  tmp/  var/  lib@    sbin@
>> dev/   home/  mnt/    proc/  run/   sys/  usr/  bin@  lib64@
>> -------------------------------------------------------------------------------------
>>
>> But now booting but hanging:
>>
>> #systemd-nspawn -bD /var/lib/machines/enl
>> Spawning container enl on /var/lib/machines/enl.
>> Press ^] three times within 1s to kill container.
>> systemd 219 running in system mode. (+PAM +AUDIT +SELINUX +IMA
>> -APPARMOR +SMACK +SYSVINIT +UTMP +LIBCRYPTSETUP +GCRYPT +GNUTLS +ACL
>> +XZ -LZ4 +SECCOMP +BLKID +ELFUTILS +KMOD +IDN)
>> Detected virtualization 'systemd-nspawn'.
>> Detected architecture 'x86-64'.
>> Running with unpopulated /etc.
>>
>> Welcome to Fedora 22 (Twenty Two)!
>>
>> Initializing machine ID from random generator.
>> Populated /etc with preset unit settings.
>> Unit etc.mount is bound to inactive unit dev-sdb1.device. Stopping, too.
>> Unit var.mount is bound to inactive unit dev-sdb1.device. Stopping, too.
>> Cannot add dependency job for unit display-manager.service, ignoring:
>> Unit display-manager.service failed to load: No such file or
>> directory.
>> Startup finished in 51ms.
>>
>> ------------------------------------------------------------------------
>>
>> Maybe my btrfs story ?
>> > Unit etc.mount is bound to inactive unit dev-sdb1.device. Stopping, too.
>> > Unit var.mount is bound to inactive unit dev-sdb1.device. Stopping, too.
>> Will investigate.
>>
>> --bind was certainly the most easy trick once the raw image is downloaded.
>
> Hmm, my guess is that you somehow lost the /etc and /var directories
> half way, probably because of the weird mounting your are
> doing. --bind= should normally be recursive, but maybe this didn't work?

root@hortensia ➤➤ lib/machines # ls poppy/etc
lvm/    systemd/  group    ld.so.cache  machine-id     os-release@  resolv.conf
pam.d/  udev/     gshadow  localtime@   nsswitch.conf  passwd       shadow

/etc seems correctly populated. Container boot but can't have login
prompt. It hangs with no possibility to enter anything.

> Lennart
>
> --
> Lennart Poettering, Red Hat



-- 

google.com/+arnaudgabourygabx

Re: [systemd-devel] install Fedora systemd-nspawn container on btrfs

[arnaud gaboury]

On Thu, Apr 23, 2015 at 7:21 PM, arnaud gaboury
<arnaud.gaboury@gmail.com> wrote:
> On Thu, Apr 23, 2015 at 7:12 PM, Lennart Poettering
> <lennart@poettering.net> wrote:
>> On Thu, 23.04.15 19:00, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:
>>
>>> On Thu, Apr 23, 2015 at 4:47 PM, Lennart Poettering
>>> <lennart@poettering.net> wrote:
>>> > On Thu, 23.04.15 14:18, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:
>>> >
>>> >> > Pick one:
>>> >> >
>>> >> > a) download the raw image and use that, but it will be a loopback file
>>> >> > with its own file system inside
>>> >> >
>>> >> > or:
>>> >> >
>>> >> > b) do the dnf/yum install root thing, and install it into a directory
>>> >> > tree.
>>> >>
>>> >> I installed yum package on Arch but couldn't manage to do the install.
>>> >>
>>> >> # yum -y --releasever=22 --nogpg --installroot=/var/lib/machines/enl
>>> >> --disablerepo='*' --enablerepo=fedora install systemd passwd dnf
>>> >> fedora-release-server
>>> >> Error getting repository data for fedora, repository not found
>>> >> # yum repolist all
>>> >> repolist: 0
>>> >>
>>> >> In fact, /etc/yum/repos.d is empty, so I am not surprised.
>>> >
>>> > Of course, you could use the .raw image, mount the extern btrfs volume
>>> > into it via nspawn's --bind= switch, then use yum inside of that
>>> > container to install into the btrfs volume. Then get rid of the .raw
>>> > image again, and you still have the btrfs volume that should be
>>> > bootable.
>>> >
>>> > A bit complex, but you almost were there already... ;-)
>>>
>>> Wunderbach
>>>
>>> # systemd-nspawn -M Fedora-Cloud-Base-22_Beta-20150415.x86_64.raw --
>>> bind=/var/lib/machines/enl:/mnt
>>> [root@Fedora-Cloud-Base-22_Beta-20150415 ~]#dnf -y --releasever=22
>>> --nogpg --installroot=/mnt
>>>  --disablerepo='*' --enablerepo=fedora install systemd passwd dnf
>>> ................
>>> Complete!
>>>
>>> -------------------------------------------------------------------------
>>> $ ls /var/lib/machines/enl
>>> boot/  etc/   media/  opt/   root/  srv/  tmp/  var/  lib@    sbin@
>>> dev/   home/  mnt/    proc/  run/   sys/  usr/  bin@  lib64@
>>> -------------------------------------------------------------------------------------
>>>
>>> But now booting but hanging:
>>>
>>> #systemd-nspawn -bD /var/lib/machines/enl
>>> Spawning container enl on /var/lib/machines/enl.
>>> Press ^] three times within 1s to kill container.
>>> systemd 219 running in system mode. (+PAM +AUDIT +SELINUX +IMA
>>> -APPARMOR +SMACK +SYSVINIT +UTMP +LIBCRYPTSETUP +GCRYPT +GNUTLS +ACL
>>> +XZ -LZ4 +SECCOMP +BLKID +ELFUTILS +KMOD +IDN)
>>> Detected virtualization 'systemd-nspawn'.
>>> Detected architecture 'x86-64'.
>>> Running with unpopulated /etc.
>>>
>>> Welcome to Fedora 22 (Twenty Two)!
>>>
>>> Initializing machine ID from random generator.
>>> Populated /etc with preset unit settings.
>>> Unit etc.mount is bound to inactive unit dev-sdb1.device. Stopping, too.
>>> Unit var.mount is bound to inactive unit dev-sdb1.device. Stopping, too.
>>> Cannot add dependency job for unit display-manager.service, ignoring:
>>> Unit display-manager.service failed to load: No such file or
>>> directory.
>>> Startup finished in 51ms.
>>>
>>> ------------------------------------------------------------------------
>>>
>>> Maybe my btrfs story ?
>>> > Unit etc.mount is bound to inactive unit dev-sdb1.device. Stopping, too.
>>> > Unit var.mount is bound to inactive unit dev-sdb1.device. Stopping, too.
>>> Will investigate.
>>>
>>> --bind was certainly the most easy trick once the raw image is downloaded.
>>
>> Hmm, my guess is that you somehow lost the /etc and /var directories
>> half way, probably because of the weird mounting your are
>> doing. --bind= should normally be recursive, but maybe this didn't work?
>
> root@hortensia ➤➤ lib/machines # ls poppy/etc
> lvm/    systemd/  group    ld.so.cache  machine-id     os-release@  resolv.conf
> pam.d/  udev/     gshadow  localtime@   nsswitch.conf  passwd       shadow
>
> /etc seems correctly populated. Container boot but can't have login
> prompt. It hangs with no possibility to enter anything.

Btrfs mess.

When in /var/lib/machines/poppy:

root@hortensia ➤➤ machines/poppy # btrfs subvolume list .
ID 266 gen 98 top level 5 path rootvol
ID 268 gen 100 top level 5 path var
ID 269 gen 101 top level 5 path etc
ID 271 gen 72 top level 266 path var/lib/machines
ID 272 gen 77 top level 268 path var/tmp
ID 273 gen 77 top level 268 path var/lib/machines

Anyone from the Btrfs ML to help ?



>
>> Lennart
>>
>> --
>> Lennart Poettering, Red Hat
>
>
>
> --
>
> google.com/+arnaudgabourygabx



-- 

google.com/+arnaudgabourygabx

Re: [systemd-devel] install Fedora systemd-nspawn container on btrfs

[Lennart Poettering]

On Thu, 23.04.15 19:29, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:

> When in /var/lib/machines/poppy:
> 
> root@hortensia ➤➤ machines/poppy # btrfs subvolume list .
> ID 266 gen 98 top level 5 path rootvol
> ID 268 gen 100 top level 5 path var
> ID 269 gen 101 top level 5 path etc
> ID 271 gen 72 top level 266 path var/lib/machines
> ID 272 gen 77 top level 268 path var/tmp
> ID 273 gen 77 top level 268 path var/lib/machines
> 
> Anyone from the Btrfs ML to help ?

Note that systemd-tmpfiles will create /var/tmp, and /var/lib/machines
as subvolumes these days, if they are missing and on btrfs.

Lennart

-- 
Lennart Poettering, Red Hat

Re: [systemd-devel] install Fedora systemd-nspawn container on btrfs

[arnaud gaboury]

On Thu, Apr 23, 2015 at 7:36 PM, Lennart Poettering
<lennart@poettering.net> wrote:
> On Thu, 23.04.15 19:29, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:
>
>> When in /var/lib/machines/poppy:
>>
>> root@hortensia ➤➤ machines/poppy # btrfs subvolume list .
>> ID 266 gen 98 top level 5 path rootvol
>> ID 268 gen 100 top level 5 path var
>> ID 269 gen 101 top level 5 path etc
>> ID 271 gen 72 top level 266 path var/lib/machines
>> ID 272 gen 77 top level 268 path var/tmp
>> ID 273 gen 77 top level 268 path var/lib/machines
>>
>> Anyone from the Btrfs ML to help ?
>
> Note that systemd-tmpfiles will create /var/tmp, and /var/lib/machines
> as subvolumes these days, if they are missing and on btrfs.
>

OK, I understand these new subvol

> Lennart
>
> --
> Lennart Poettering, Red Hat



-- 

google.com/+arnaudgabourygabx

Re: [systemd-devel] install Fedora systemd-nspawn container on btrfs

[arnaud gaboury]

On Thu, Apr 23, 2015 at 7:37 PM, arnaud gaboury
<arnaud.gaboury@gmail.com> wrote:
> On Thu, Apr 23, 2015 at 7:36 PM, Lennart Poettering
> <lennart@poettering.net> wrote:
>> On Thu, 23.04.15 19:29, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:
>>
>>> When in /var/lib/machines/poppy:
>>>
>>> root@hortensia ➤➤ machines/poppy # btrfs subvolume list .
>>> ID 266 gen 98 top level 5 path rootvol
>>> ID 268 gen 100 top level 5 path var
>>> ID 269 gen 101 top level 5 path etc
>>> ID 271 gen 72 top level 266 path var/lib/machines
>>> ID 272 gen 77 top level 268 path var/tmp
>>> ID 273 gen 77 top level 268 path var/lib/machines
>>>
>>> Anyone from the Btrfs ML to help ?
>>
>> Note that systemd-tmpfiles will create /var/tmp, and /var/lib/machines
>> as subvolumes these days, if they are missing and on btrfs.
>>
>
> OK, I understand these new subvol

After cleaning everything, new install. Fedora 22 container boot but
then hangs: no login prompt.

I did this:
enablerepo=fedora install systemd passwd dnf fedora-release-server

Maybe did I forgot one essential repo?

I will switch to Fedora ML now.
TY Lennart for your hints.

>
>> Lennart
>>
>> --
>> Lennart Poettering, Red Hat
>
>
>
> --
>
> google.com/+arnaudgabourygabx



-- 

google.com/+arnaudgabourygabx

Re: [systemd-devel] install Fedora systemd-nspawn container on btrfs

[arnaud gaboury]

On Fri, Apr 24, 2015 at 8:14 AM, arnaud gaboury
<arnaud.gaboury@gmail.com> wrote:
> On Thu, Apr 23, 2015 at 7:37 PM, arnaud gaboury
> <arnaud.gaboury@gmail.com> wrote:
>> On Thu, Apr 23, 2015 at 7:36 PM, Lennart Poettering
>> <lennart@poettering.net> wrote:
>>> On Thu, 23.04.15 19:29, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:
>>>
>>>> When in /var/lib/machines/poppy:
>>>>
>>>> root@hortensia ➤➤ machines/poppy # btrfs subvolume list .
>>>> ID 266 gen 98 top level 5 path rootvol
>>>> ID 268 gen 100 top level 5 path var
>>>> ID 269 gen 101 top level 5 path etc
>>>> ID 271 gen 72 top level 266 path var/lib/machines
>>>> ID 272 gen 77 top level 268 path var/tmp
>>>> ID 273 gen 77 top level 268 path var/lib/machines
>>>>
>>>> Anyone from the Btrfs ML to help ?
>>>
>>> Note that systemd-tmpfiles will create /var/tmp, and /var/lib/machines
>>> as subvolumes these days, if they are missing and on btrfs.
>>>
>>
>> OK, I understand these new subvol
>
> After cleaning everything, new install. Fedora 22 container boot but
> then hangs: no login prompt.
>
> I did this:
> enablerepo=fedora install systemd passwd dnf fedora-release-server
>
> Maybe did I forgot one essential repo?

EDIT : if it can help:

gabx@hortensia ➤➤ ~ % systemd-cgls
├─1 /usr/lib/systemd/systemd
├─machine.slice
│ └─machine-poppy.scope
│   ├─5654 /usr/lib/systemd/systemd
│   └─system.slice
│     ├─lvm2-lvmetad.service
│     │ └─5678 /usr/sbin/lvmetad -f
│     ├─systemd-journald.service
│     │ └─5676 /usr/lib/systemd/systemd-journald
│     └─dm-event.service
│       └─5668 /usr/sbin/dmeventd -f


>
> I will switch to Fedora ML now.
> TY Lennart for your hints.
>
>>
>>> Lennart
>>>
>>> --
>>> Lennart Poettering, Red Hat
>>
>>
>>
>> --
>>
>> google.com/+arnaudgabourygabx
>
>
>
> --
>
> google.com/+arnaudgabourygabx



-- 

google.com/+arnaudgabourygabx

Re: [systemd-devel] install Fedora systemd-nspawn container on btrfs

[arnaud gaboury]

On Fri, Apr 24, 2015 at 8:16 AM, arnaud gaboury
<arnaud.gaboury@gmail.com> wrote:
> On Fri, Apr 24, 2015 at 8:14 AM, arnaud gaboury
> <arnaud.gaboury@gmail.com> wrote:
>> On Thu, Apr 23, 2015 at 7:37 PM, arnaud gaboury
>> <arnaud.gaboury@gmail.com> wrote:
>>> On Thu, Apr 23, 2015 at 7:36 PM, Lennart Poettering
>>> <lennart@poettering.net> wrote:
>>>> On Thu, 23.04.15 19:29, arnaud gaboury (arnaud.gaboury@gmail.com) wrote:
>>>>
>>>>> When in /var/lib/machines/poppy:
>>>>>
>>>>> root@hortensia ➤➤ machines/poppy # btrfs subvolume list .
>>>>> ID 266 gen 98 top level 5 path rootvol
>>>>> ID 268 gen 100 top level 5 path var
>>>>> ID 269 gen 101 top level 5 path etc
>>>>> ID 271 gen 72 top level 266 path var/lib/machines
>>>>> ID 272 gen 77 top level 268 path var/tmp
>>>>> ID 273 gen 77 top level 268 path var/lib/machines
>>>>>
>>>>> Anyone from the Btrfs ML to help ?
>>>>
>>>> Note that systemd-tmpfiles will create /var/tmp, and /var/lib/machines
>>>> as subvolumes these days, if they are missing and on btrfs.
>>>>
>>>
>>> OK, I understand these new subvol
>>
>> After cleaning everything, new install. Fedora 22 container boot but
>> then hangs: no login prompt.
>>
>> I did this:
>> enablerepo=fedora install systemd passwd dnf fedora-release-server
>>
>> Maybe did I forgot one essential repo?
>
> EDIT : if it can help:
>
> gabx@hortensia ➤➤ ~ % systemd-cgls
> ├─1 /usr/lib/systemd/systemd
> ├─machine.slice
> │ └─machine-poppy.scope
> │   ├─5654 /usr/lib/systemd/systemd
> │   └─system.slice
> │     ├─lvm2-lvmetad.service
> │     │ └─5678 /usr/sbin/lvmetad -f
> │     ├─systemd-journald.service
> │     │ └─5676 /usr/lib/systemd/systemd-journald
> │     └─dm-event.service
> │       └─5668 /usr/sbin/dmeventd -f

Issue solved:
Fedora release 22 (Twenty Two)
Kernel 3.19.3-3-apparmor on an x86_64 (console)

poppy login:

What I did: umount my two btrfs subvol
 # umount /var/lib/machines/poppy/etc
 # umount /var/lib/machines/poppy/var

then restart the container.

There is something weird with my btrfs settings.


>
>
>>
>> I will switch to Fedora ML now.
>> TY Lennart for your hints.
>>
>>>
>>>> Lennart
>>>>
>>>> --
>>>> Lennart Poettering, Red Hat
>>>
>>>
>>>
>>> --
>>>
>>> google.com/+arnaudgabourygabx
>>
>>
>>
>> --
>>
>> google.com/+arnaudgabourygabx
>
>
>
> --
>
> google.com/+arnaudgabourygabx



-- 

google.com/+arnaudgabourygabx