Buffalo NAS-Central Forums

Hi all, My first post here...

I'm in the process of upgrading the disks on my LS-QVL. (SSH enabled)

I'm moving from 4*1TB to 4*4TB disks.

My question is about Chunk size for the md2 raid 5.

Firstly, could anyone here do a mdadm on their raid 5 md2 and let me know the chunk size please?

The reason I ask is mine is in the final stage of reshaping (3->4) and is showing an 8K chunk size which seems to me too small.

I have a feeling that this chunk size may have been as a result of the way I migrated my data. What I did was this...

1. First copied my data from the original 3TB raid 5 to one of the new 4TB disks on USB
2. Broke the raid 5 and put another of my new 4TB disks in the now vacant slot.
3. cp the data from the now degraded 3TB raid to the internal 4TB disk - I then had two copies of my data
4. Removed the remaining 3*1TB disks from the QVL chassis.
5. Put the remaining two 4TB into two of the vacant slots and created a new raid 1 using the GUI
6. cp the data from the single 4TB disk to the new 4TB raid 1
7. Removed the single 4TB disk and wiped the partition table of that and the 4TB USB disk.

I now had a secure copy of my original data on a raid 1 - capacity 4TB

8. Place the remaining two wiped disks in the chassis.
9. Used the GUI RMM function to migrate the raid1 to raid5 and add one of the new disks

At this point I noticed that the new raid 5 had the 8k chunk size

10. I'm now in the process of adding the fourth drive to the raid 5 using the GUI RMM.

I'll do some performance tests when it has finished the reshape but I'd really like to know what chunk size other folks have.

Many thanks,

Ian

Many thanks for that. Mine currently looks like...

Personalities : [linear] [raid0] [raid1] [raid6] [raid5] [raid4]
md2 : active raid5 sdd6[3] sdc6[2] sda6[0] sdb6[1]
7782872816 blocks super 1.2 level 5, 8k chunk, algorithm 2 [4/4] [UUUU]
[=========>...........] reshape = 49.0% (1908512672/3891436408) finish=1851.0min speed=17854K/sec

md1 : active raid1 sdd2[7] sdc2[4] sda2[5] sdb2[6]
4999156 blocks super 1.2 [4/4] [UUUU]

md10 : active raid1 sdd5[7] sdc5[4] sda5[5] sdb5[6]
1000436 blocks super 1.2 [4/4] [UUUU]

md0 : active raid1 sdd1[3] sdc1[0] sda1[1] sdb1[2]
1000384 blocks [4/4] [UUUU]

unused devices: <none>

Looks like I'll have to change the Chunk size - another 2 days.
I'll check the performance before I do.

Cheers,

Ian

For future reference (should this happen to anyone else) this is what I discovered...

When I tried to change the chunk size on the array I got an error message indicating that the number of disk blocks was not divisible by my new chunk size of 512.

It seems that when I created the raid1 it used all blocks available in partition 6 (since the chunk is not used in raid1).
Then when I migrated the raid1 to raid5 the maximum chunk size that would fit was 8k.

Note:- create a raid5 with 3 disks and then add the additional disk.