RaidZ2 setup questions - different size disks

[alttab]

Hi there,

Currently I am configuring my nas\virtualbox server and I want to make a RaidZ2 pool containing 7 disks.

Disks I have in my NAS right now:
5x 3TB disk (WD and Seagate)
2x 1TB disk

Disk I have outside my nas, in my PC:
1x 3TB disk (containing data)

My plan was to make a RaidZ2 pool of all 7 disks in the NAS --> copy over the data of the 3TB disk of my pc to the pool --> and afterwards replacing one of the 1tb disks in my NAS with this 3TB disk and let it rebuild.
Eventually this would result in a RAIDZ2 pool of 6x3TB and 1x1TB in my nas.
I got stuck on creating the pool, I got the error 'raidz contains devices of different sizes' , so I have to use the force option - this got me thinking

(the reason I want this small 1 TB drive to stay in this pool, is the fact that I want to upgrade it to a larger disk later on , I don't care about this 1TB space honestly )

I think I need to get my plans doublechecked before I do something stupid and loose my data
Some questions;

- what are the consequences of mixing drive sizes in a raidz2 pool?
for instance in the case above, what happens if 2 3TB drives would fail, and the 1TB disk stays alive, will I be able to rebuild?

and ;
- what if I only put 6x 3TB in a zraid2 pool, and in the future, I replace one of these 3tb disks with a 5TB disk. (resulting in 5x3TB + 1x 5TB) , what happens then if the large disk dies together with a small disk?
still able to rebuild?

- any other advice on this case?

Thanks!

Edit: found the answers allready. All 3TB drives will only deliver 1TB as long as this smaller drive is in the pool.
Now i have to decide; Buy another 3TB disk and make a 7x3TB Raidz2 pool, or make a 6x3TB Raidz2 pool.

[slaycock]

I read on this forum that for raidzfs2 there are performance gains from keeping the data drives as power of 2 so that the total number of drives in the array is 2^n+2 (2 data +2 parity, 4 data + 2 parity, 8 data + 2 parity etc for n=1,2,3).

[alttab]

Something went wrong with the replacement of the drive.

Anyone has a clue what happened here? Something to do with the IBM 1015 controller? (IT Mode)

This is what i did:

- Made a RAIDZ2 pool , 5x 3TB, 1x 1TB
- copied the data over to the pool, now i want to change the 1TB with a fresh 3TB.
- took 1TB offline (da4.nop)
- shutdown server, replace the 1TB drive with the 3TB drive
- boot server: disk management: drive ids are messed up, not 1 as i would expect, but 3 drives have a new id! checked serial numbers... yes they got swapped
another drive took the place of da4.nop , the replacement drive got da5.nop , but its on the same SATA cable as the old da4.nop ...
- so, i am unable to do the replace command on da4.nop, because of the fact that another drive took this place and this drive is online so i get an error..
- put the 1TB back, and all is fine again ...

I choose another path and now i will make a fresh pool of 6 3tb disks , so i won't have to replace the disk anymore.
But still i want to know why all these disk ids got swapped when i replaced only 1 disk.
Someone knows?

[raulfg3]

first eliminate .nop to easy references, see this post: viewtopic.php?f=66&t=2982

once done repeat your step until step 5, at this point you have shift disk names, (searh forum for shif disk syndrome) well you need to sync webGUI and real world, go to disk and use new button "Clear config & import Disk", this sync your NAS WebGUI and your real Disk names.

next go to Disks|ZFS|Configuration|Sync and use Sync button to sync detected(real) and current(nas4free webgui saved) pool.

http://wiki.nas4free.org/doku.php?id=do ... n_detected

http://wiki.nas4free.org/doku.php?id=do ... ynchronize


once done you are in condition to execute replace command with confidence.

some examples: viewtopic.php?f=66&t=4210

http://people.freebsd.org/~pjd/misc/zfs ... eplace.swf