You are not logged in.
I started out installing my root partition on a RAID-0 array with two drives.
The two drives are 180 and 120 GB respectively and I have a 100GB+100GB array setup
Now I added yet another 2 drives to my laptop with the DVD-ROM caddy adapter and an m-Sata card.
I would like to migrate my installation so I can have a 4 way array.
How can I migrate?
I believe the following is the right way:
Make an empty partition, copy everything in / to that and boot into that install by specifying the root= option in grub
Then remake the RAID array and copy everything back
Will this work?
Thanks in advance
Last edited by rep_movsd (2015-02-26 19:19:17)
Offline
So you want to have RAID0 with 4 disks? Wouldn't something like https://wiki.archlinux.org/index.php/RA … ID0_Arrays work?
Offline
WOW I had no idea RAID 0 could grow!
Will it restripe the existing data across the newly added drives?
Last edited by rep_movsd (2015-02-26 06:59:40)
Offline
Seems like this wont work:
Offline
Have you actually tried it? That bug report is old.
It should restripe data.
In case it isn't clear, you should probably use "mdadm --grow /dev/md0 --raid-devices=4 --add /dev/sdc1 /dev/sdd1".
Offline
I made a test array with two small partitions, copied some data onto it and tried the following:
rep ~ $ cat /proc/mdstat
Personalities : [raid0] [raid6] [raid5] [raid4]
md2 : active raid0 sdc3[1] sda4[0]
33718272 blocks super 1.2 64k chunks
md0 : active raid0 sda1[0] sdc1[1]
203776 blocks super 1.2 512k chunks
md1 : active raid0 sda2[0] sdc2[1]
200055296 blocks super 1.2 512k chunks
unused devices: <none>
rep ~ $ sudo mdadm --grow /dev/md2 --level=0 --raid-devices=3 --add /dev/sdd1
mdadm: /dev/md2: could not set level to raid4
rep ~ $
What now?
FYI it is not mounted either
Last edited by rep_movsd (2015-02-26 09:17:03)
Offline
What if you remove "--level=0"?
Offline
rep ~ $ sudo mdadm --grow /dev/md2 --raid-devices=3 --add /dev/sdd1
mdadm: /dev/md2: could not set level to raid4
This happens
dmesg shows :
[10231.922029] md/raid:md2: cannot takeover raid0 with more than one zone.
[10231.922038] md: md2: raid4 would not accept array
[10231.923433] md2: p1
Last edited by rep_movsd (2015-02-26 10:03:02)
Offline
Are the partitions in md2 the same size? http://forums.gentoo.org/viewtopic-p-71 … ml#7158464
Offline
Let me retry making the test once more from scratch
Offline
Thanks!
Remaking the partitions to the exact same size worked!
Now I shall make a backup of my / filesystem and try adding to it
Offline
4 drive raid0? You do have backups, right?
Offline
I'm using SSDs from Intel, Crucial and Samsung
I've never experienced sudden failure on my SSDs without SMART warnings since I started using them in 2009. My 40 GB Intel from then is still working great on a desktop system (FWIW a WD Raptor from 2006 is still working perfectly, no SMART warnings, it's booted almost daily since then and was my primary system for a few years - must be some sort of record!)
All my personal data is backed up, work stuff is on assembla, so I can take the risk!
The main advantage of this striping comes when you work with multiple VMs - Suspending and resuming them is super fast with RAID0
Offline
Further progress and further stumbling block
So I started by making a full rsync backup of the / partition and booted into that successfully
After that I discovered that the initial two raid0 partitions differed in size, so I deleted and recreated that array, now with three partitions.
I updated /etc/mdadm.conf and ran mkinitcpio to create a fresh initrd
Now rsynced the backup onto the new raid0 array
Tried to reboot, but grub couldnt find it's theme files etc and gave me a text menu. The backup partition still boots
Edited the grub config, it had the old GUID in the "search" statements
I changed that GUID to the new one from /etc/mdadm.conf
Again rsynced the changes to the new root
Rebooted again...
Now GRUB still fails to find the theme files
When I select the old entry, the kernel loads, it gets as far as "Started /dev/md1 etc etc"
After that it says "fsck.e2 failed for /dev/md1, Error 2 : not found" or something to that effect and gives me a root shell, but for some reason the tty doesnt work and I cant type anything.
Note that my /boot partition is /dev/md0
Whats to be done next?
Offline
I have a feeling I messed up something :
rep ~ $ lsblk -f
NAME FSTYPE LABEL UUID MOUNTPOINT
sda
|-sda1 linux_raid_member archiso:0 514bcc23-3a29-8779-235e-390743d2e5e9
| `-md0 ext4 b2b35d72-71b7-4558-beb0-eceacc59294f /boot
|-sda2 linux_raid_member archaic:1 238d7fab-995b-c7b8-6fd9-b884dd5974e8
| `-md1
| `-md1p1 ext4 raid_root e2f763af-0c17-40af-bcd8-6be6bfed00da /mnt/new_raid
|-sda3 ext4 c84f122a-c7ca-4716-a9cb-19b7e34fb819
`-sda4 ext4 backup dd0a5e00-c231-411d-ba8b-27006d40c8fc /
sdb
`-sdb1 ext4 data b7c6a22b-8734-48d6-bb31-35367ac5b7b9 /mnt/500g
sdc
|-sdc1 linux_raid_member archiso:0 514bcc23-3a29-8779-235e-390743d2e5e9
| `-md0 ext4 b2b35d72-71b7-4558-beb0-eceacc59294f /boot
`-sdc2 linux_raid_member archaic:1 238d7fab-995b-c7b8-6fd9-b884dd5974e8
`-md1
`-md1p1 ext4 raid_root e2f763af-0c17-40af-bcd8-6be6bfed00da /mnt/new_raid
sdd
|-sdd1 ntfs 30F93E06015AD5DF /mnt/test
`-sdd2 linux_raid_member archaic:1 238d7fab-995b-c7b8-6fd9-b884dd5974e8
`-md1
`-md1p1 ext4 raid_root e2f763af-0c17-40af-bcd8-6be6bfed00da /mnt/new_raid
rep ~ $
In this new configuration, the filesystem is on /dev/md1p1 instead of md1 directly
Earlier I am sure it was directly on /dev/md1
I used gparted to do that - perhaps I should have run mkfs directly on /dev/md1 rather than let gparted make another partition table within it
Offline
I redid the operation using mkfs and updated grub
Now it works great!
Read speed up from ~400 MB/s to 600 MB/s
Offline
How do you mark a topic solved? I'm sure there used to be an option somewhere, cant find it though!
[Solved]
Offline
Edit the first post.
Offline