6 drives, 2x3tb 4x2tb, 3tb show as 2tb WTF

mdgm wrote:
I recently expanded my Ultra 6 volume. I had 6x1.5TB disks installed and I'm using X-RAID2 dual-redundancy. I had to install 4x3TB drives before expansion took place.

Thats an initial setup where you need the 4 disks to start RAID 6 with mdadm software. However this doesnt cover the expansion topic where you only need 3.

You will need to look at his logs when he adds a 3tb drive.

Md3 will be raid 6 with [UUU] - The raw disk space will be 1 TB extra

Each X represent 1 TB of raw disk space

XX
XXXXX
XXXXX

Currently here is what is going on (from here on out each X represents 1 TB of raw disk space and each O represents 1 TB of redundant space)

OO
OOXXXX
OOXXXX

[Raw disk space used (not including overhead and not including kilo to kibi conversions): 8TB and redundancy space will be: 3 TB]


The second he adds a third drive it will work

OOX
OOXXXXX
OOXXXXX

[Raw disk space used (not including overhead and not including kilo to kibi conversions): 11TB and redundancy space will be: 6 TB]

Although just having three 3TB disks will not give you RAID 6.

However in the setup of going from 3,3,2,2,2,2 to 3,3,3,2,2,2 I believe it will mdadm "cat /proc/mdstat" will say raid 6 and it will show you [UUU] and the extra TB will be used

Any how What it comes down to is this Add another 3 TB drive or add 2 more 3 TB drive. Lets see when it actually expands the volume using some of the extra md3 space (the extra space from the 3tb drives)

##############

I understand that if you start for example with

two disks X tb each and then add another X tb drive then it can be RAID 5 and then add another X tb disk it can become RAID 6.

Then if you add 3 disks of Y size. Y being X+a. You will be able to see the +a being used for volume after the 3rd disk

NOTE: () Parenthesis around the Redundancy

================================

(Y)XXXX raid 6

[Equals]

(a)
(X)XXXX

all the X portions make md2, the extra "a" from the Y disk makes md3, Ys extra "a" is wasted

================================

Same with the next setup

(YY)XXXX raid 6, both Ys extra double a are wasted

(aa)
(XX)XXXX

Both those "a" make the md3 but they cant be "raid6"ed yet by mdadm so its useless - in a way

================================

With next add ition mdadm will allow the dual redundancy to work

(aa)a
(YY)YXXXX raid 6

================================

This is from experience at looking through various setups like this

I will verify if im correct or wrong with L3 tommorow

I suppose I should specify the constraints first.

I would like to have X-RAID2 dual redundancy and have a mix of 3TB and 2TB disks available for use. My desire at first was to first use my 2TB disks since I've had those the longest but after reading through the responses, it sounds like I should be using at least 4x3TB and only then use the 2TB disks. With the recent sales on 3TB disks, I could load up all 6 slots with 3TB disks and then save the 2TB disks for an offline backup.

kossboss wrote:
mdgm wrote: I recently expanded my Ultra 6 volume. I had 6x1.5TB disks installed and I'm using X-RAID2 dual-redundancy. I had to install 4x3TB drives before expansion took place. Thats an initial setup where you need the 4 disks to start RAID 6 with mdadm software. However this doesnt cover the expansion topic where you only need 3.

????? I last did a factory default back on 4.2.12 (yes, that's right 4.2.12) in July 2010 with 6x1.5TB disks installed and configured the NAS to use X-RAID2 dual-redundancy. I added the 3TB disks late last year to expand my volume.

kevikev wrote:

I suppose I should specify the constraints first. I would like to have X-RAID2 dual redundancy and have a mix of 3TB and 2TB disks available for use. My desire at first was to first use my 2TB disks since I've had those the longest but after reading through the responses, it sounds like I should be using at least 4x3TB and only then use the 2TB disks. With the recent sales on 3TB disks, I could load up all 6 slots with 3TB disks and then save the 2TB disks for an offline backup.

4x3TB+2x2TB or 6x3TB are best. I guess 5x3TB is also reasonable, leaving 1 slot open for future expansion.

Well You were right "ITS IMPOSSIBLE TO MAKE RAID 6 WITH 3 MEMBERS"
I tested it

┌─[✗]─[deb32]─[/vol/raid]
└──╼ mdadm -Cv -c64 -l6 -n3 /dev/md10 /dev/loop1 /dev/loop2 /dev/loop3
mdadm: at least 4 raid-devices needed for level 6

But raid 5 can be made with 2

Lol :-)