Discussion:
Unsuccessful root installation with ubuntu 12.04 and separate ext4 installation
Ben Jamin
2013-04-23 17:09:04 UTC
Permalink
Hello,

I rent a small kimsufi server at ovh.
The server is supplied with 2 500GB drives
I do not have access to the server console and i haven't had success to
remote boot a live cd in order to complete steps described here
: https://github.com/zfsonlinux/pkg-zfs/wiki/HOWTO-install-Ubuntu-to-a-Native-ZFS-Root-Filesystem

So i have tried a variant of this procedure.
Here are the exact steps i followed :
- install kimsufi server with the following partition scheme
- /boot in mdadm mirror : so /dev/sda1 and /dev/sdb1 go in /dev/md1
ext4 of 500MB
- / in mdadm mirror : so /dev/sda2 and /dev/sdb2 go in /dev/md2 ext4 of
2GB
- swap because no choice with ovh
- /* INSTALL KERNEL STANDARD to replace custom ovh kernel incompatible with
module loading */
passwd
apt-get -y install linux-image-server && mkdir ~/ovh.d && mv
/etc/grub.d/06_OVHkernel ~/ovh.d && ls ~/ovh.d && update-grub && reboot
- /* INSTALL HEADERS */
uname -a
apt-get -y install linux-headers-`uname -r` dkms build-essential
reboot
- /* INSTALL of locales en_US and replacement of en_GB by default */
locale
nano /etc/default/locale
/* CONTENU LOCALE */
LANG=en_US.UTF-8
LC_MESSAGES=POSIX
/* CONTENU LOCALE */

update-locale LANG=en_US.UTF-8 LC_MESSAGES=POSIX
locale -a
locale-gen -a
reboot
locale
/* LOCALE OUTPUT */
LANG=en_US.UTF-8
LANGUAGE=
LC_CTYPE="en_US.UTF-8"
LC_NUMERIC="en_US.UTF-8"
LC_TIME="en_US.UTF-8"
LC_COLLATE="en_US.UTF-8"
LC_MONETARY="en_US.UTF-8"
LC_MESSAGES=POSIX
LC_PAPER="en_US.UTF-8"
LC_NAME="en_US.UTF-8"
LC_ADDRESS="en_US.UTF-8"
LC_TELEPHONE="en_US.UTF-8"
LC_MEASUREMENT="en_US.UTF-8"
LC_IDENTIFICATION="en_US.UTF-8"
LC_ALL=
/* LOCALE OUTPUT */

- /* REMOVAL OF SWAP (IT SHOULD GO ON A RAID0 ZVOL LATER) AND partitioning
swapoff -a
cfdisk
cfdisk /dev/sdb
partprobe
- /* comment out swap reference in fstab */
nano /etc/fstab
reboot
- /* INSTALL ZFS */
apt-get -y install python-software-properties
apt-add-repository --yes ppa:zfs-native/stable
apt-get update
apt-get -y install debootstrap ubuntu-zfs
modprobe zfs
==> ok
ls -la /dev/disk/by-id
fdisk -l
- /* POOL AND DATASET CREATION */
zpool create -o ashift=9 rpool mirror
/dev/disk/by-id/ata-ST3500418AS_6VM80B1D-part3
/dev/disk/by-id/ata-TOSHIBA_DT01ACA050_233SPANPS-part3
zfs create rpool/ROOT
zfs create rpool/ROOT/ubuntu-1
df
reboot
df
partprobe
df
fdisk -l
- /* LABEL /boot / DON'T KNOW IF NECESSARY */
e2label /dev/md1 "/boot"
blkid
- /* SET ZFS PROPERTIES FOR BOOT AND MOUNT */
zfs umount -a
zfs set mountpoint=/ rpool/ROOT/ubuntu-1
zpool set bootfs=rpool/ROOT/ubuntu-1 rpool
zpool export rpool
zpool import -d /dev/disk/by-id -R /mnt rpool
df
- /* LET'S DEBOOTSTRAP ON NEW ZFS DATASET */
debootstrap precise /mnt
- /* LET'S GIVE ACCESS TO VIRTUAL FILE SYSTEM */
/* SINCE I WISH TO KEEP THE EXISTING UBUNTU INSTALLATION I KEEP /BOOT AND
BIND IT */
mount --bind /boot /mnt/boot
mount --bind /dev /mnt/dev
mount --bind /proc /mnt/proc
mount --bind /sys /mnt/sys
- /* EDIT FSTAB TO ONLY KEEP REFERENCE TO /BOOT */
nano /mnt/etc/fstab
/* CONTENT */
# UNCONFIGURED FSTAB FOR BASE SYSTEM
/dev/md1 /boot ext4 errors=remount-ro,relatime 0 1
/* CONTENT */

- /* COPY NECESSARY FILES FOR NETWORK AND APT */
cp /etc/hostname /mnt/etc/
cp /etc/hosts /mnt/etc/
cp /etc/network/interfaces /mnt/etc/network/
cp /etc/apt/sources.list /mnt/etc/apt/
- /* LET'S GO */
chroot /mnt /bin/bash --login
locale-gen en_US.UTF-8
apt-get update
apt-get -y install ubuntu-minimal python-software-properties
apt-add-repository --yes ppa:zfs-native/stable
apt-add-repository --yes ppa:zfs-native/grub
apt-get update
uname -a
apt-get -y install --no-install-recommends linux-image-`uname -r`
linux-headers-`uname -r`
apt-get -y install linux-firmware
apt-get -y install ubuntu-zfs
apt-get -y install grub2-common grub-pc
apt-get -y install zfs-initramfs
apt-get -y install openssh-server openssh-client
apt-get -y dist-upgrade
apt-get -y install pciutils
passwd root
grub-probe /
==> zfs
ls /boot/grub/zfs*
==> ok
grub-install $(readlink -f /dev/disk/by-id/ata-ST3500418AS_6VM80B1D)
grub-install $(readlink -f
/dev/disk/by-id/ata-TOSHIBA_DT01ACA050_233SPANPS)
update-initramfs -c -k all
update-grub
grub-install $(readlink -f /dev/disk/by-id/ata-ST3500418AS_6VM80B1D)
grub-install $(readlink -f
/dev/disk/by-id/ata-TOSHIBA_DT01ACA050_233SPANPS)
grep boot=zfs /boot/grub/grub.cfg
==> linux /vmlinuz-3.2.0-40-generic root=/dev/sda3 ro boot=zfs $bootfs
linux /vmlinuz-3.2.0-40-generic root=/dev/sda3 ro single
nomodeset boot=zfs $bootfs
/* NOTE : i don't have any reference to rpool/ROOT/ubuntu-1 here what's
going wrong ? */
exit
umount /mnt/boot/
umount /mnt/dev
umount /mnt/proc
umount /mnt/sys
zfs umount -a
/* NEED TO KILL SO PROCESS THERE FOR SUCCESSFUL UMOUNT */
zpool export rpool
reboot
At reboot the server comes back with ping but no ssh, no webmin even if
everything is configured. Since i have no access to the console, the
debugging is really hard and syslog don't tell me anything.
When i come to this point, i can only select to netboot on specific ovh
environment chroot the linux installation on /dev/md2 reinstall and update
grub and reboot to go back on the linux on the first partition. But from
ovh environment, i cannot load zfs modules.
I've tried in zfs chroot to reinstall grub specifying zfs-grub and grub-pc.
They are the good version with +zfs

I would really be happy running zfs on this plateform. Do you have any
suggestions ? of corrections or debugging ?

Thanks in advance for sharing your knowledge.
Ben Jamin
2013-04-23 17:13:41 UTC
Permalink
Just an update here is the complete grub.cfg generated.

Benjamin
Ben Jamin
2013-04-24 13:25:25 UTC
Permalink
Hello,

Sorry for insistence but i'm really stuck on this installation i don't see
how to get further. Everything seems correct except grub.cfg generation
which outputs :
grep boot=zfs /boot/grub/grub.cfg
==> linux /vmlinuz-3.2.0-40-generic root=/dev/sda3 ro boot=zfs $bootfs
linux /vmlinuz-3.2.0-40-generic root=/dev/sda3 ro single
nomodeset boot=zfs $bootfs

It's the good grub version since it adds correctly "boot=zfs $bootfs" but
it do not set the approppriate root.

Do you have any advise to give me to go further in debugging ? i've tried
to enable bootlogd but it seems no supported anymore.

I would really appreciate your help.
I could after create a tutorial for other kimsufi users at ovh and we are
many i think.

Regards.

Benji
Swâmi Petaramesh
2013-04-24 13:46:48 UTC
Permalink
Post by Ben Jamin
Sorry for insistence but i'm really stuck on this installation i don't see
how to get further. Everything seems correct except grub.cfg generation
grep boot=zfs /boot/grub/grub.cfg
==> linux /vmlinuz-3.2.0-40-generic root=/dev/sda3 ro boot=zfs $bootfs
linux /vmlinuz-3.2.0-40-generic root=/dev/sda3 ro single
nomodeset boot=zfs $bootfs
It's the good grub version since it adds correctly "boot=zfs $bootfs" but
it do not set the approppriate root.
Mine acted up the same way, I resorted to manually editing the grub.cfg and
remove the «root=/dev/blah» (My case is worse as I use RAIDZ and got
"root=/dev/sda2 root=/dev/sdb2»).

If you just remove this, it should work.

Kind regards.
--
Swâmi Petaramesh <swami-***@public.gmane.org> http://petaramesh.org PGP 9076E32E
Ben Jamin
2013-04-24 14:07:34 UTC
Permalink
Hello,

Thanks for your answer.
menuentry 'Ubuntu, with Linux 3.2.0-40-generic' --class ubuntu --class gnu-linux --class gnu --class os {
recordfail
gfxmode $linux_gfx_mode
insmod gzio
insmod raid
insmod mdraid09
insmod part_msdos
insmod part_msdos
insmod ext2
set root='(md1)'
search --no-floppy --fs-uuid --set=root 7a2b1ce0-0f9a-4551-8b50-e03b4741a5bb
insmod zfsinfo
zfs-bootfs ($root) bootfs
linux /vmlinuz-3.2.0-40-generic root=/dev/sda3 ro boot=zfs $bootfs
initrd /initrd.img-3.2.0-40-generic
}
menuentry 'Ubuntu, with Linux 3.2.0-40-generic' --class ubuntu --class gnu-linux --class gnu --class os {
recordfail
gfxmode $linux_gfx_mode
insmod gzio
insmod raid
insmod mdraid09
insmod part_msdos
insmod part_msdos
insmod ext2
set root='(md1)'
search --no-floppy --fs-uuid --set=root 7a2b1ce0-0f9a-4551-8b50-e03b4741a5bb
insmod zfsinfo
zfs-bootfs ($root) bootfs
linux /vmlinuz-3.2.0-40-generic ro boot=zfs $bootfs
initrd /initrd.img-3.2.0-40-generic
}
Or anything else ?
Did you notice i had

linux /vmlinuz-3.2.0-40-generic ro boot=zfs $bootfs
initrd /initrd.img-3.2.0-40-generic

Instead of :

linux /ROOT/ubuntu-1/@/boot/vmlinuz-3.2.0-40-generic ro boot=zfs $bootfs
initrd /ROOT/ubuntu-1/@/boot/initrd.img-3.2.0-40-generic


Thanks in advance for your support.

Benji
Post by Ben Jamin
Sorry for insistence but i'm really stuck on this installation i don't
see
Post by Ben Jamin
how to get further. Everything seems correct except grub.cfg generation
grep boot=zfs /boot/grub/grub.cfg
==> linux /vmlinuz-3.2.0-40-generic root=/dev/sda3 ro boot=zfs
$bootfs
Post by Ben Jamin
linux /vmlinuz-3.2.0-40-generic root=/dev/sda3 ro single
nomodeset boot=zfs $bootfs
It's the good grub version since it adds correctly "boot=zfs $bootfs"
but
Post by Ben Jamin
it do not set the approppriate root.
Mine acted up the same way, I resorted to manually editing the grub.cfg and
remove the «root=/dev/blah» (My case is worse as I use RAIDZ and got
"root=/dev/sda2 root=/dev/sdb2»).
If you just remove this, it should work.
Kind regards.
--
http://petaramesh.org PGP 9076E32E
Swâmi Petaramesh
2013-04-24 14:20:19 UTC
Permalink
Hi,
Post by Ben Jamin
Thanks for your answer.
[...]
Post by Ben Jamin
set root='(md1)'
search --no-floppy --fs-uuid --set=root
7a2b1ce0-0f9a-4551-8b50-e03b4741a5bb insmod zfsinfo
zfs-bootfs ($root) bootfs
linux /vmlinuz-3.2.0-40-generic ro boot=zfs $bootfs
initrd /initrd.img-3.2.0-40-generic
[...]
Post by Ben Jamin
Or anything else ?
Did you notice i had
linux /vmlinuz-3.2.0-40-generic ro boot=zfs $bootfs
initrd /initrd.img-3.2.0-40-generic
If I understand well from this, you have your /boot on an ext4 MD-RAID (md1)
and then your root FS is ZFS RAIDZ on sda3 (and sdb3 ?)

Then it's normal that you have «linux /vmlinuz...» as it's relative to your
/boot (md1) and the kernel is in the root of this md, isn't it ?

(grub understand the «set root=blah» as defining the device where the linux
kernel is located, not the device where the future filesystem root is, which
grub doesn't really need or care about...)

If I were you, I would try the way you understood, and should you have
problems with the "zfs-bootfs ($root) bootfs» line, then remove it also, drop
the $bootfs parameter and use:

linux /vmlinuz-3.2.0-40-generic ro boot=zfs rpool=YOUR_ZFS_POOL_NAME_HERE

Kind regards.
--
Swâmi Petaramesh <swami-***@public.gmane.org> http://petaramesh.org PGP 9076E32E
Ben Jamin
2013-04-24 14:45:17 UTC
Permalink
Post by Ben Jamin
Hi,
Post by Ben Jamin
Thanks for your answer.
[...]
Post by Ben Jamin
set root='(md1)'
search --no-floppy --fs-uuid --set=root
7a2b1ce0-0f9a-4551-8b50-e03b4741a5bb insmod zfsinfo
zfs-bootfs ($root) bootfs
linux /vmlinuz-3.2.0-40-generic ro boot=zfs $bootfs
initrd /initrd.img-3.2.0-40-generic
[...]
Post by Ben Jamin
Or anything else ?
Did you notice i had
linux /vmlinuz-3.2.0-40-generic ro boot=zfs $bootfs
initrd /initrd.img-3.2.0-40-generic
boot=zfs $bootfs
If I understand well from this, you have your /boot on an ext4 MD-RAID (md1)
and then your root FS is ZFS RAIDZ on sda3 (and sdb3 ?)
You're correct my /boot is on /dev/md1 (mdadm mirror)
I have an installation of ubuntu on /dev/md2 (which i wish to keep as a
rescue OS)
I have created rpool/ROOT/ubuntu-1 on /dev/sda3 and /dev/sdb3 (by id)

Then it's normal that you have «linux /vmlinuz...» as it's relative to
Post by Ben Jamin
your
/boot (md1) and the kernel is in the root of this md, isn't it ?
***@ks352670:/# ll /boot/
total 32664
drwxr-xr-x 4 root root 1024 Apr 24 13:41 ./
drwxr-xr-x 23 root root 26 Apr 22 18:07 ../
-rw-r--r-- 1 root root 794933 Mar 25 22:02 abi-3.2.0-40-generic
-rw-r--r-- 1 1002 1002 6134704 Apr 2 2012
bzImage-3.2.13-xxxx-grs-ipv6-64
-rw-r--r-- 1 root root 140586 Mar 25 22:02 config-3.2.0-40-generic
drwxr-xr-x 3 root root 8192 Apr 24 13:41 grub/
-rw-r--r-- 1 root root 16271962 Apr 24 13:41 initrd.img-3.2.0-40-generic
drwx------ 2 root root 12288 Apr 21 17:48 lost+found/
-rw------- 1 root root 2890703 Mar 25 22:02 System.map-3.2.0-40-generic
-rw-r--r-- 1 1002 1002 2206409 Apr 2 2012
System.map-3.2.13-xxxx-grs-ipv6-64
-rw------- 1 root root 4974672 Mar 25 22:02 vmlinuz-3.2.0-40-generic

and i have symlink from root
***@ks352670:/# ll /
[...]
lrwxrwxrwx 1 root root 32 Apr 21 18:42 initrd.img ->
boot/initrd.img-3.2.0-40-generic
[...]
lrwxrwxrwx 1 root root 29 Apr 21 18:42 vmlinuz ->
boot/vmlinuz-3.2.0-40-generic
[...]
Post by Ben Jamin
(grub understand the «set root=blah» as defining the device where the
linux
kernel is located, not the device where the future filesystem root is, which
grub doesn't really need or care about...)
If I were you, I would try the way you understood, and should you have
problems with the "zfs-bootfs ($root) bootfs» line, then remove it also,
drop
linux /vmlinuz-3.2.0-40-generic ro boot=zfs rpool=YOUR_ZFS_POOL_NAME_HERE
Kind regards.
--
http://petaramesh.org PGP 9076E32E
I'll give it a try when i'm back home in few minutes
Ben Jamin
2013-04-24 16:52:11 UTC
Permalink
Hello,

I finally found a way to access remotely the server using vkvm (which for
some reason didn't want to work the previous days)

Here is what i get with the original grub

<Loading Image...>
It seems that bootfs property is not correctly read.
Anyway i rebooted and edited grub entry to this :

<Loading Image...>
And here is where it brings me

<Loading Image...>
I then tried the following step of the howto and the command executed
successfully.
But after I go this

<Loading Image...>
I really feel i'm close to the goal but stuck here.

Thanks for your support

Regards

Benji
Swâmi Petaramesh
2013-04-24 18:03:49 UTC
Permalink
Hi Benjamin,

Following your posts I took a closer look at my grub.cfg generation, and found
a much more elegant way to have it work properly (and auto-generate) by just
editing /etc/default/grub and putting into it the following values (for my
system, adapt for yours as you see fit...) :

# This is where my ZFS root FS resides (mine is on LUKS-encrypted LVM):
GRUB_DEVICE="/dev/mapper/VG1-UZ1"
GRUB_DEVICE_UUID=""
GRUB_FS="zfs"
GRUB_CMDLINE_LINUX="boot=zfs rpool=UZ1"

And then "update-grub" does the trick.

Should it still complain that it doesn't find the device for "/", then edit
/usr/sbin/grub-mkconfig, and hardcode GRUB_DEVICE= GRUB_DEVICE_UUID= and
GRUB_FS= (avoiding the call to ${grub_probe}) (circa lines 140-160)

Should work ;-)
--
Swâmi Petaramesh <swami-***@public.gmane.org> http://petaramesh.org PGP 9076E32E

Aussi longtemps nous sommes dans l'accès, nous ne parvenons pas jusque-là.
-- Maître Eckhart - Sermon 71 (XIVe s.)
Ben Jamin
2013-04-24 18:11:41 UTC
Permalink
Thanks a lot for the advice. I'm terribly sorry but i cannot find how to
map this to my system
GRUB_DEVICE="/dev/md0" ==> my /boot ???
GRUB_DEVICE_UUID=""
GRUB_FS="zfs"
GRUB_CMDLINE_LINUX="boot=zfs rpool=rpool" OR "boot=zfs
rpool=rpool/ROOT/ubuntu-1"

Sorry for my bad understanding.
Post by Swâmi Petaramesh
Hi Benjamin,
Following your posts I took a closer look at my grub.cfg generation, and found
a much more elegant way to have it work properly (and auto-generate) by just
editing /etc/default/grub and putting into it the following values (for my
GRUB_DEVICE="/dev/mapper/VG1-UZ1"
GRUB_DEVICE_UUID=""
GRUB_FS="zfs"
GRUB_CMDLINE_LINUX="boot=zfs rpool=UZ1"
And then "update-grub" does the trick.
Should it still complain that it doesn't find the device for "/", then edit
/usr/sbin/grub-mkconfig, and hardcode GRUB_DEVICE= GRUB_DEVICE_UUID= and
GRUB_FS= (avoiding the call to ${grub_probe}) (circa lines 140-160)
Should work ;-)
--
http://petaramesh.org PGP 9076E32E
Aussi longtemps nous sommes dans l'accÚs, nous ne parvenons pas jusque-là.
-- Maître Eckhart - Sermon 71 (XIVe s.)
Swâmi Petaramesh
2013-04-24 19:19:44 UTC
Permalink
Post by Ben Jamin
Thanks a lot for the advice. I'm terribly sorry but i cannot find how to
map this to my system
GRUB_DEVICE="/dev/md0" ==> my /boot ???
I would say yes :-)
Post by Ben Jamin
GRUB_DEVICE_UUID=""
GRUB_FS="zfs"
GRUB_CMDLINE_LINUX="boot=zfs rpool=rpool" OR "boot=zfs
rpool=rpool/ROOT/ubuntu-1"
Juste put "rpool=rpool" here, but make sure you have set the "bootfs" property
for the zfs pool to the complete root fs path. i.e. on another of my machines
(this one running ArchLinux) :

# zpool get bootfs
NAME PROPERTY VALUE SOURCE
FZ1 bootfs FZ1/ARCH/ROOTFS local

Kind regards.
--
Swâmi Petaramesh <swami-***@public.gmane.org> http://petaramesh.org PGP 9076E32E
Marius Bjørnstad
2013-04-24 18:23:13 UTC
Permalink
I haven't followed the thread, but I have seen "filesystem '-' cannot be
mounted" before. It happens because "-" is the default value for bootfs.
You could try using "zpool set bootfs=rpool/ROOT/ubuntu-1 /root rpool".
I never got "bootfs" to work on my scientific linux system though, using
a "root=" kernel parameter like the other reply suggested

--Marius
Post by Ben Jamin
Hello,
I finally found a way to access remotely the server using vkvm (which
for some reason didn't want to work the previous days)
Here is what i get with the original grub
<https://lh5.googleusercontent.com/-BQLafNYevFY/UXgKpDrUz0I/AAAAAAAACZU/XZiBFY-V790/s1600/boot_kimsufi_zfs_1.jpg>
It seems that bootfs property is not correctly read.
<https://lh3.googleusercontent.com/-N5jNvZhwm6E/UXgLR2DHu7I/AAAAAAAACZc/R0y5EJlhRvc/s1600/boot_kimsufi_zfs_2.jpg>
And here is where it brings me
<https://lh5.googleusercontent.com/-K4YdrKbLYL8/UXgL70gBA5I/AAAAAAAACZk/EpIDlK0vrYE/s1600/boot_kimsufi_zfs_3.jpg>
I then tried the following step of the howto and the command executed
successfully.
But after I go this
<https://lh5.googleusercontent.com/-FcJjtaRBru0/UXgNiLvsVLI/AAAAAAAACZw/OsWjxSsUwA0/s1600/boot_kimsufi_zfs_4.jpg>
I really feel i'm close to the goal but stuck here.
Thanks for your support
Regards
Benji
Marius Bjørnstad
2013-04-24 18:35:03 UTC
Permalink
My reply was wrong in at least 2 ways, please ignore. I apologise for
any confusion and wasted time. The errors are quite strange, I haven't
seen "X cannot be mounted, unable to open dataset". You could try things
like zpool status and zfs list in BusyBox to see why your dataset is not
there. Also try zpool export and import if there is anything but ONLINE
in zpool status. If the pool becomes accessible, you could run the mount
command that's written after "Command:" in the screenshot again, and
then type exit or Ctrl-D, and maybe it will boot. That will probably not
fix it permanently if it does help.


--Marius
Post by Marius Bjørnstad
I haven't followed the thread, but I have seen "filesystem '-' cannot
be mounted" before. It happens because "-" is the default value for
bootfs. You could try using "zpool set bootfs=rpool/ROOT/ubuntu-1
/root rpool". I never got "bootfs" to work on my scientific linux
system though, using a "root=" kernel parameter like the other reply
suggested
--Marius
Post by Ben Jamin
Hello,
I finally found a way to access remotely the server using vkvm (which
for some reason didn't want to work the previous days)
Here is what i get with the original grub
<https://lh5.googleusercontent.com/-BQLafNYevFY/UXgKpDrUz0I/AAAAAAAACZU/XZiBFY-V790/s1600/boot_kimsufi_zfs_1.jpg>
It seems that bootfs property is not correctly read.
<https://lh3.googleusercontent.com/-N5jNvZhwm6E/UXgLR2DHu7I/AAAAAAAACZc/R0y5EJlhRvc/s1600/boot_kimsufi_zfs_2.jpg>
And here is where it brings me
<https://lh5.googleusercontent.com/-K4YdrKbLYL8/UXgL70gBA5I/AAAAAAAACZk/EpIDlK0vrYE/s1600/boot_kimsufi_zfs_3.jpg>
I then tried the following step of the howto and the command executed
successfully.
But after I go this
<https://lh5.googleusercontent.com/-FcJjtaRBru0/UXgNiLvsVLI/AAAAAAAACZw/OsWjxSsUwA0/s1600/boot_kimsufi_zfs_4.jpg>
I really feel i'm close to the goal but stuck here.
Thanks for your support
Regards
Benji
Ben Jamin
2013-04-24 18:41:39 UTC
Permalink
Ok we're going further. I've added the requested suggestion in
/etc/default/grub and also modified /usr/share/initramfs-tools/scripts/zfs
to add a sleep 5 before the mount line.
As a result, the pool imports automatically but I'm still blocked just
after with this :

<Loading Image...>
What does those udevd errors mean ?
timeout 'write_cd_rules'
timeout: killing 'write_cd_rules'

Any ideas ?
I feel i'm close to the goal.

Regards,

Benji
Swâmi Petaramesh
2013-04-24 19:21:22 UTC
Permalink
Post by Ben Jamin
What does those udevd errors mean ?
timeout 'write_cd_rules'
timeout: killing 'write_cd_rules'
Looks like a cd-rom drive identification issue, I guess this shouldn't be
blocking though...
--
Swâmi Petaramesh <swami-***@public.gmane.org> http://petaramesh.org PGP 9076E32E

Tout ce qu'on peut acquérir n'est qu'une chose de plus à perdre.
-- Choke
Ben Jamin
2013-04-24 19:48:10 UTC
Permalink
you were right it was a cd-rom identification issue. I guess it's been
caused by my debootstrap install. I've taken back the files from
/etc/udev/rules.d of the original install and the message disappeared.
Now i'm blocked on the previous message saying that :
the disk drive for /tmp is not ready yet or not present.

It should be there since it's part of the rpool. Can this be due to the
fact i have no swap in my install ?


Regards.
Post by Swâmi Petaramesh
Post by Ben Jamin
What does those udevd errors mean ?
timeout 'write_cd_rules'
timeout: killing 'write_cd_rules'
Looks like a cd-rom drive identification issue, I guess this shouldn't be
blocking though...
--
http://petaramesh.org PGP 9076E32E
Tout ce qu'on peut acquérir n'est qu'une chose de plus à perdre.
-- Choke
Ben Jamin
2013-04-25 16:59:12 UTC
Permalink
This post might be inappropriate. Click to display it.
Swâmi Petaramesh
2013-04-25 17:58:31 UTC
Permalink
Post by Ben Jamin
've also tried to add vdev_id support following the procedure of the faq,
udevd[xxxx]: timeout: killing /dev/xxx for any kind of hardware.
Do you have any idea again ?
This issue is most probably not ZFS related, so it is probably not appropriate
to discuss it on the ZFS mailing-list...
--
Swâmi Petaramesh <swami-***@public.gmane.org> http://petaramesh.org PGP 9076E32E

Motivation (n) Salaire virtuel destiné à faire économiser de l'argent aux
employeurs.
Ben Jamin
2013-04-25 18:06:56 UTC
Permalink
Are you sure since vdev_id is part of zfs library?
Regards

Benji
Ben Jamin
2013-04-25 18:21:56 UTC
Permalink
https://lh5.googleusercontent.com/-FcJjtaRBru0/UXgNiLvsVLI/AAAAAAAACZw/OsWjxSsUwA0/s1600/boot_kimsufi_zfs_4.jpg

The errors I get now are the than the one of this screenshot to be more precise
Ben Jamin
2013-04-25 20:49:32 UTC
Permalink
Finally i did it
df
Filesystem 1K-blocks Used Available Use% Mounted on
rootfs 196115456 1328256 194787200 1% /
udev 1009240 4 1009236 1% /dev
tmpfs 407864 272 407592 1% /run
rpool/ROOT/ubuntu-1 196115456 1328256 194787200 1% /
none 5120 4 5116 1% /run/lock
none 1019660 0 1019660 0% /run/shm
/dev/md1 494756 49023 420188 11% /boot
rpool 194787328 128 194787200 1% /rpool
rpool/ROOT 194787328 128 194787200 1% /rpool/ROOT

I think that the install with debootstrat was not enough on my kimsufi
hardware i tried another way.
Instead of :
debootstrap precise /mnt
I did
rsync -axHAX / /mnt/

As we say simple is beautiful.
Now i have a few more things to do
- create ZVOL for swap
- try snapshot on rpool and see if update-grup takes them into account
- ...

Une bonne chose de faite

Benji

Loading...