ONTAP Discussions

RAID group not configured correctly?

scott_lau
7,995 Views

I have an aggregate that was set with a RAID group size of 13 and 3 disks were added prior to the RAID group increase to 16. How can i clean this up to increase performance properly?

Thanks!

11 REPLIES 11

glynbowden
7,960 Views

Migration is really your only option now. If you had enough disks to hold the data thats on there you would have to make a new aggr, copy the data onto it (couple of options for that). Recreate this aggr correctly and then copy back.

The copying options could include a straight NDMP copy, snapmirror or if you can get the licence you might be able to get a syncmirror running. I don't recall if the two mirrors need to be excatly symetrical though, so you would need to check that too.

Other than that, I'm afraid there aren't so many options. If it is causing real issues, you could have a word with your account manager and maybe lease a shelf or something to perform the migration.

Sorry its not better news.

Glyn

glynbowden
7,959 Views

Just as a quick note, I did a quick experiment by flipping the RAID type from RAID_DP to RAID4 just to see if it would force an aggr layout change. As expected, it did, but only within the RAID groups, and as there is no way to remove disks from an aggr (just like the original trad vols), then I'm afraid its migration time. When adding disks to aggregates its always worth previewing the RAID layout before proceeding. This can limit the amount of surprises that crop up later.

eng-nf01> aggr add aggr_test 3
Note: preparing to add 1 data disk and 2 parity disks.
Continue? ([y]es, [n]o, or [p]review RAID layout) p
The RAID group configuration will change as follows:
RAID Group                Current               NEW
----------                -------               ----
/aggr_test/plex0/rg0      5 disks               5 disks
/aggr_test/plex0/rg1                            3 disks

Continue? ([y]es, [n]o, or [p]review RAID layout) y
Addition of 3 disks to the aggregate has completed.

Glyn

jb2cool01
7,960 Views

As far as i know, you cannot without destroying the aggregate and then recreating it from scratch. You might also be able to upgrade out of this issue by adding more disks to fill in the gaps but i'm not 100% sure that'll work.

leericmarvin
7,958 Views

OK, I have same problem as original poster.   In my case, I have a brand new NetApp 2240 with 24ea 1TB disks.  With no data on it other than the vol0. 

The RAID groups were set to 11, so when I added disks, it created a new RAID group.  I later discovered the RAID groups can be adjusted up to 28 disks.

So, I have at this moment two RG's with 2 parity, 9 data and 1 spare on each.   I'd really like to have one RG with 2 parity 20 data and 2 spare. 

Other than increasing the data drive count by two, Is having one RG better then two?

If one RG is better, what is best way to do this?  Do I put the two spare disk in there own small aggregate, "copy/migrate" the vol0 to it and delete the "bad" aggregate? 

It looks like the system boots on a solid state drive,  and puts config files in the vol0?  Can I just delete the aggregate and reboot system and rerun setup command (which I am getting use to by now), or is there critical files there?

Other option, what does it take to just reinstall the NetApp OS? Amongest all the documentation and paperwork that came with the system, there isn't any CD's or software.

Download to laptop and putty the OS in?

Any ideas here appreicated.

LeEric

aborzenkov
7,958 Views

You can go via both routes - create small aggregate or completely recreate root aggregate. The former has advantage of preserving all existing settings. If you decide to reinstall, do not forget to capture licenses (or better make sure you can fetch them from support.netapp.com). You will need to reenter them. But in this case you do not actually “reinstall Data ONTAP” - you just recreate root aggregate.

Data ONTAP 8.x binaries are completely located on internal boot device (it is USB Flash Module in newer hardware). And if you ever need to replace boot device, you use netboot with image downloaded from support.netapp.com.

leericmarvin
7,958 Views

Thanks for the response!

OK, so if I get this right, I should go ahead and create a small aggregate of two drives (from the two spares remaining) one for data and other for parity. I then copy the root volume to this two drive aggregate. After that I boot on the new volume? Then delete the large aggregate.  recreate that aggregate the correct way. Copy the root vol back to this and then delete the small aggregate leaving those drives as spare.

LeEric

aborzenkov
7,958 Views

Yes. Do not forget to mark new volume as root (vol options new_volume_on_small_aggregate root). And you will need to mark volume on new big aggregate as root again when you copy data back.

leericmarvin
7,960 Views

OK, I have a problem.  Using NetApp OnCommand System Manager 2.0 I attempt to create a new aggregate of just 2 disks. But I get warning that I need to have a spare disk. I try again with one disk, but get prompted to add a 2nd disk for the RAID4.   If I try and up the disk count, wraps right back to refusing on account of no spare being left.   

I get the feeling I'll need to reinstall the OS.

leericmarvin
7,958 Views

Is it possible to change the large aggr to RAID4 causing 2 disks to become spares? Or will they turn into data disk?

ismopuuronen
5,946 Views

Hi,

Did you tried it?
I think it's possible to change raid type from raid-dp to raid 4 causing one parity disk per rg to drop to the spare pool.
If you are having raid 4 in use, you will need at least on spare disk. If you are having only raid dp in use, you don't have to had any spare disks.
>options raid.min_spare_count <value>

Br.
Ismo.

leericmarvin
5,947 Views

I have been in communication with my netapp resellers engineer.

I am at home at the moment and not at work where I can see my notes, so from memory......

He point me to a boot option which allowed me to re-initialize the Netapp back to it's original state.

I then tried to grow the default Aggr0 and RAID from 1 disk and two parity drives to 20/2 and it would not go. Gave some error when I tried to raise RAID size passed 20.

The engineer checked his sources and came back stating the RAID can't be bigger then 20 with the parity drives in that count. in other words 18/2.

Also recommended was making the two RAIDs in also two different Aggr.

So: 

Aggr0  RAID 1/2

Arrg1  RAID 18/2

leaves one for spare.

Thanks for everyone's contributions with this.  I am going to proceed with the above.

Public