I’m planning to migrate from ubuntu server to open suse server
I’m testing the migration in a virtual machine and I’m having problems with mdadm
Every time Open suse boots I need to run mdadm --assemble /dev/md0 and activate the volume group vgchange -ay VGNAME
During the boot time boot.md and mdadmd are running but (startup scripts only scan) I believe mdmad is not running because needs to be assemble every time.
Any idea?
Thanks
qsummon wrote:
>
> I’m planning to migrate from ubuntu server to open suse server
> I’m testing the migration in a virtual machine and I’m having problems
> with mdadm Every time Open suse boots I need to run
> mdadm --assemble /dev/md0 and activate the volume group vgchange -ay
> VGNAME
> During the boot time boot.md and mdadmd are running but (startup
> scripts only scan) I believe mdmad is not running because needs to be
> assemble every time.
> Any idea?
Try creating /etc/mdadm.conf
–
Per Jessen, Zürich (10.6°C)
http://en.opensuse.org/User:pjessen
Thanks Jessen
Already got one, here is a copy
Any idea why I need to assemble the raid every time I boot
Thanks
nfs-test01:~ # cat /etc/mdadm.conf
DEVICE partitions
ARRAY /dev/md0 level=raid5 auto=md metadata=00.90 UUID=5d1bb32b:da6964c8:46a7ba41:b057fcdb
CREATE owner=root group=disk mode=0660 auto=yes
HOMEHOST <nfs-test01>
MAILADDR root
On 2010-10-10 08:06, qsummon wrote:
>
> I’m planning to migrate from ubuntu server to open suse server
> I’m testing the migration in a virtual machine and I’m having problems
> with mdadm
> Every time Open suse boots I need to run mdadm --assemble /dev/md0 and
> activate the volume group vgchange -ay VGNAME
The partition type is fd?
–
Cheers / Saludos,
Carlos E. R.
(from 11.2 x86_64 “Emerald” at Telcontar)
qsummon wrote:
>
> Thanks Jessen
> Already got one, here is a copy
> Any idea why I need to assemble the raid every time I boot
>
Are the partition types set up as 0xFD ?
–
Per Jessen, Zürich (9.9°C)
http://en.opensuse.org/User:pjessen
No partition type. I have created the Vg and lvm directly to the disk and was working with no issues and Ubuntu is recognizing it at boot time. I beliieve is not necessary to create a partition for it
Here is the output from fdisk, lvdisplay and vgdisplay from the virtual machine I’m using for the test (I have the same in my physical production server)
FDISK
Disk /dev/md0: 3219 MB, 3219652608 bytes
2 heads, 4 sectors/track, 786048 cylinders
Units = cylinders of 8 * 512 = 4096 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 524288 bytes / 1572864 bytes
Disk identifier: 0x00000000
Disk /dev/md0 doesn’t contain a valid partition table
LVDISPLAY
— Logical volume —
LV Name /dev/vgtest/lvtest
VG Name vgtest
LV UUID JEZ2eO-hPid-DS73-VZj7-ir7w-kJdp-80vBIO
LV Write Access read/write
LV Status available
LV Size 3.00 GiB
Current LE 767
Segments 1
Allocation inherit
Read ahead sectors auto
VGDISPLAY
nfs-test01:~ # vgdisplay -v vgtest
Using volume group(s) on command line
Finding volume group “vgtest”
— Volume group —
VG Name vgtest
System ID
Format lvm2
Metadata Areas 1
Metadata Sequence No 8
VG Access read/write
VG Status resizable
MAX LV 0
Cur LV 1
Open LV 0
Max PV 0
Cur PV 1
Act PV 1
VG Size 3.00 GiB
PE Size 4.00 MiB
Total PE 767
Alloc PE / Size 767 / 3.00 GiB
Free PE / Size 0 / 0
VG UUID VfCfkF-8SwO-fn1A-3OWr-B1yC-MIJ3-zjfV1T
— Logical volume —
LV Name /dev/vgtest/lvtest
VG Name vgtest
LV UUID JEZ2eO-hPid-DS73-VZj7-ir7w-kJdp-80vBIO
LV Write Access read/write
LV Status available
LV Size 3.00 GiB
Current LE 767
Segments 1
Allocation inherit
Read ahead sectors auto
— Physical volumes —
PV Name /dev/md0
PV UUID OnNyZr-5Csu-Wq3z-Ofb9-C1dX-27Qo-xMt9cG
PV Status allocatable
Total PE / Free PE 767 / 0
I went trough the following exercise in the open suse server (Virtual Machine)
I delete the 3 drive I was using for the mdadm raid (This raid was created on ubuntu server)
I create 3 new drives and create a new mdmad raid
mdadm --create /dev/md0 --chunk=128 --level=5 --raid-devices=3 /dev/sdb /dev/sdc /dev/sdd
My first assumption was the old raid was not working at the boot time because was created in a diffreret OS, but with this NEW raid still not working at boot time.
I have to manually run mdadm --assemble -s in order to see the disk
Just a quick reminder of my project.
I have a physical server I want to migrate from ubuntu to open suse
I’m testing the migration on a virtual machine I build with the same parameters as the physical one (just smaller drives)
I assign the 3 disk to the open suse server and mdmadm is not workling at the boot time.
Thank you very much
Marcelo
Why not use the YaST partitioner to create the RAID filesystem when installing? It should result in the correct initrd sequence to detect and assemble the RAID. Also it should put the required modules in the initrd.
What does your GRUB kernel line look like? Hopefully something like this?
kernel /boot/vmlinuz-2.6.31.12-0.1-default root=/dev/md0 blah blah
I haven’t checked, but I think that root=/dev/md0 will cause the initrd to assemble the RAID for the root filesystem so that it can be mounted and the boot sequence continues with a valid /etc. After that /etc/mdadm.conf should direct the assembly of further filesystems. As you realise, the RAID for / has to be a special case.
The boot and OS partitions (root/etc) are not an mdadm partition. I’m booting from one ssd.
I want to be able to recognize one existing raid 5 I already create on Ubuntu, not to create a new one
How I get access to yast partitioner? I’m just seeing the general settings under the /etc/sysconfig editor
I’m running Open Suse 11.3 as a server
Thanks
qsummon wrote:
>
> No partition type. I have created the Vg and lvm directly to the disk
> and was working with no issues and Ubuntu is recognizing it at boot
> time. I beliieve is not necessary to create a partition for it
Can we see the output from “cat /proc/mdstat” please?
I’m not concerned with your LVM setup, that’s not where the problem is.
–
Per Jessen, Zürich (8.6°C)
http://en.opensuse.org/User:pjessen
On 2010-10-11 16:06, qsummon wrote:
>
> I went trough the following exercise in the open suse server (Virtual
> Machine)
> I delete the 3 drive I was using for the mdadm raid (This raid was
> created on ubuntu server)
> I create 3 new drives and create a new mdmad raid
> mdadm --create /dev/md0 --chunk=128 --level=5 --raid-devices=3 /dev/sdb
> /dev/sdc /dev/sdd
>
> My first assumption was the old raid was not working at the boot time
> because was created in a diffreret OS, but with this NEW raid still not
> working at boot time.
> I have to manually run mdadm --assemble -s in order to see the disk
>
> Just a quick reminder of my project.
> I have a physical server I want to migrate from ubuntu to open suse
> I’m testing the migration on a virtual machine I build with the same
> parameters as the physical one (just smaller drives)
> I assign the 3 disk to the open suse server and mdmadm is not workling
> at the boot time.
You will have to expand the explanation of the disk layout as seen by the guest. I thought you had a
software raid made of partitioned disks. In that setup, partitions have to be of type FD for linux
to attempt to assemble the array during boot.
Now it seems there is an LVM setup on the hard disks, and raid on top the LVM. Or is it the other
way round? Please explain in detail.
I don’t have this clear, and anyway, as I don’t like LVM and I don’t understand it, I withdraw.
–
Cheers / Saludos,
Carlos E. R.
(from 11.2 x86_64 “Emerald” at Telcontar)
this is the output before manually assemble the raid
nfs-test01:~ # more /proc/mdstat
Personalities : [linear]
unused devices: <none>
This is the outout after mdadm
nfs-test01:/etc # mdadm --assemble /dev/md0
mdadm: /dev/md0 has been started with 4 drives.
nfs-test01:/etc # cat /proc/mdstat
Personalities : [linear] [raid6] [raid5] [raid4]
md0 : active raid5 sdb[0] sde[3] sdd[2] sdc[1]
3144192 blocks level 5, 512k chunk, algorithm 2 [4/4] [UUUU]
unused devices: <none>
Thanks
I use Volume Groups and Logical volumes. I’m not using partition tables ( I believe in not needed). I got the same in all my VMs and in my physical
here is a complete layout of the disk
DISK 1
/boot (only ext3 non-lvm)
Volume group VGROOT
Logical Volumes under it
/home
/opt
/root
/swap
/tmp
/usr
/var
DISK2,3,4 (Set as Raid 5)
Volume Group VGTEST
Logical Volume lvtest
On 2010-10-11 21:06, qsummon wrote:
> I want to be able to recolonize and existing raid 5 I already create on
> Ubuntu, not to create a new one
But you would learn how yast creates it. I think it would perhaps copy /etc/mdadm.conf to initrd, or
something of the sort.
–
Cheers / Saludos,
Carlos E. R.
(from 11.2 x86_64 “Emerald” at Telcontar)
I couldn’t find yast partitioner. The only thing I’m seeing is mdadm in /etc/sysconfig Editor (in yast)
where is the partitioner?
I’m using Open Suse 11.3
YaST > System > Partitioner
and if you don’t have it, perhaps you don’t have the yast2-storage package installed?
On 2010-10-12 15:06, qsummon wrote:
>
> I couldn’t find yast partitioner. The only thing I’m seeing is mdadm in
> /etc/sysconfig Editor (in yast)
> where is the partitioner?
> I’m using Open Suse 11.3
What? Yast is the jewel of the crown, the single thing that diferentiates suse/opensuse from any
other distro.
It is a program. It is named “yast” in the menu, or call “yast2” in a root’s xterm. Then go to
section system, there is the partitioner.
But in order to learn how yast sets such a system, you would need to do this from the installer, I
think.
–
Cheers / Saludos,
Carlos E. R.
(from 11.2 x86_64 “Emerald” at Telcontar)
I’m not seeing partitioner under System
I’m only seeing
/etc/sysconfig Editor
Date and Time
Language
System services (run level)
On 2010-10-12 19:06, qsummon wrote:
>
> I’m not seeing partitioner under System
> I’m only seeing
> /etc/sysconfig Editor
> Date and Time
> Language
> System services (run level)
Then start the software manager module, search for things with “yast” in the name, and start adding
modules.
–
Cheers / Saludos,
Carlos E. R.
(from 11.2 x86_64 “Emerald” at Telcontar)
qsummon wrote:
>
> this is the output before manually assemble the raid
>
> nfs-test01:~ # more /proc/mdstat
> Personalities : [linear]
> unused devices: <none>
>
> This is the outout after mdadm
> nfs-test01:/etc # mdadm --assemble /dev/md0
> mdadm: /dev/md0 has been started with 4 drives.
>
> nfs-test01:/etc # cat /proc/mdstat
> Personalities : [linear] [raid6] [raid5] [raid4]
> md0 : active raid5 sdb[0] sde[3] sdd[2] sdc[1]
> 3144192 blocks level 5, 512k chunk, algorithm 2 [4/4] [UUUU]
>
> unused devices: <none>
>
> Thanks
Okay, no partition types needed.
I took another look at your mdadm.conf - your ARRAY lines are a little
different to what I have on a vanilla openSUSE installation:
DEVICE containers partitions
ARRAY /dev/md0 UUID=8b34420b:eafcc774:19203a7a:fb36d9f8
ARRAY /dev/md1 UUID=c7b3e5d5:f1815443:776c2c25:004bd7b2
ARRAY /dev/md2 UUID=19a4169b:6dcc4e04:776c2c25:004bd7b2
I don’t know if your extra parameters make a difference though.
–
Per Jessen, Zürich (9.4°C)
http://en.opensuse.org/User:pjessen