OVH Community, your new community space.

help with my harddisks


zimsters
18-02-2012, 10:51
Quote Originally Posted by Thelen
All sorted now? Let me know if you want to change the RAID without losing data, I've found a neat way thanks to the guy who maintains mdadm

for the most part! I think my sdc is actually a failed drive - it's showing as failed in all my info. I'll need to run a check and get ovh to fix it up if it is infact a problem. won't be messing around with raids until that's all fixed!

but it'd be sweet to know your trick around changign raid levels I guess next logical step after 5 would be 10? not sure of the performance impact - i want what's fastest obviously.

Thelen
18-02-2012, 01:35
All sorted now? Let me know if you want to change the RAID without losing data, I've found a neat way thanks to the guy who maintains mdadm

zimsters
17-02-2012, 20:11
Quote Originally Posted by Myatu
md3 is in RAID-5, which tolerates the loss of one HD (which is also how OVH could replace the HDD without affecting your data).

As for "splitting", simply remove scc/sdd from md1 and create a new volume for them (see 'man mdadm').

And checking if the partition is used as swap, just enter 'fdisk -l /dev/sdb' (or any other "/dev/sdX") and it'll show how it is partitioned
yep it's swap. thanks so much for all your help

Myatu
17-02-2012, 20:04
md3 is in RAID-5, which tolerates the loss of one HD (which is also how OVH could replace the HDD without affecting your data).

As for "splitting", simply remove scc/sdd from md1 and create a new volume for them (see 'man mdadm').

And checking if the partition is used as swap, just enter 'fdisk -l /dev/sdb' (or any other "/dev/sdX") and it'll show how it is partitioned

zimsters
16-02-2012, 19:26
Is there no risk in removing sda from the md3 volume? I guess it's 1 of the 4 drives so it should be able to tolerate that level of fault?

following that my guess is just reformat sda splitting it into sda1/2/3, add 3 back into md3.

how can i 'split' the md1 drive and 'reassign' the sdc/sdd drives to a separate mirrored volume, leaving sda1/sdb1 in the current md1 volume?

also how can i check whether sdb/c/d2 are being used for swap?

thanks so much for the help!

Myatu
16-02-2012, 19:16
Yeah, sda does appear to be under-utilized. You can remove it from the raid array, re-partition it as the other disks, and re-attach it to the raid array (both MD1 and MD3) and have it rebuilt.

Also, MD1 uses 3 disks in RAID 1 configuration (plus sda1 if you were to re-partition and re-attach it). RAID 1 is merely mirroring (so 1 disk is the original, the other 2 [3] are copies). This would OK if you need ultra fault tolerance (1 dies, you still have 2 or 3 disks with the same content to use), but otherwise, you could split this up (Ie., two separate raid-partitions).

You may not be seeing sdb2, sdc2, etc being used, as those partitions could be Swap partitions.

zimsters
16-02-2012, 12:11
hey all,

I've got one of the older kimsufis with 4 x harddisks, can't recall capacity.

I had had a drive failure some time back and ovh had replaced it. I'm just now checking status of my drives and I'm not sure if I'm taking advantage of it all.

In /dev/ I have:
sda
sdb / sdb1 / sdb2 / sdb3
sdc / sdc1 / sdc2 / sdc3
sdd / sdd1 / sdd2 / sdd3

According to cat /proc/mdstat:

MD3 (active raid 5) using:
sda[0], sdd3[3] and sdb3[1], sdc3[4]

md1 (active raid1) using:
sdb1[4], sdd1[3], sdc1[4]


Now to me that doesn't sound like I'm making use of all the redundancy / advantages I can get? For one why does sda not have any partitions, it's just sda? second, appears that sdb2, sdc2, sdd2 are not being used at all. Also given md1 is raid1 but has 3 drives - is this not a waste of space? Not that I need the storage space, but wouldn't it be better off upping this to another raid level?

Help me optimise this setup and ensure that I'm making most use of my storage and have redundancy in place please?