Re: Converting a FAS2240-2 to a disk shelf

MKELLYSCSC · ‎2016-05-03

Hi,

We have a FAS2240-2 with 24 x 900GB SAS drives running ONTAP 8.1.1P4 7-Mode.

We would like to convert it to a DS2246 disk shelf and connect it to a FAS2552 with 24 x 900GB SAS drives running CDOT 8.3.2 (no external disk shelves).

I've been reading the documentation on converting a 2240 to a disk shelf and the process seems overly complex.

Is it not possible to:

Boot the 2240 into maintenance mode and zero the disks.
Power off the 2240 and replace the controllers with IOM6 modules.
Power on the (old) 2240 and change the shelf id.
Connect it to the FAS2552 filer?

We did something similar recently when we (non-disruptively) moved 3 x 4243 disk shelves from a FAS3210 running 8.1.1P1 7-Mode to a FAS8020 with 4 existing 4243 disk shelves running CDOT 8.3.1P1.

All we had to do was zero the disks on the 3210 before cabling the shelves to the new system.

Thanks.

aborzenkov · ‎2016-05-04

Documentation describes how to convert FAS2240 into shelf while retaining data on internal disks. If you do not need data, you can simply replace controllers with IOM and connect to existing FAS. All further work (reassigning and zero disks) can be done online from this system.

Is it not possible to:
Boot the 2240 into maintenance mode and zero the disks.

No really; disk zero_spares command is not available in maintenance mode (I do not think you can even destroy aggregate in maintenance mode before you can start zeroing disks). What you can do is to reassign ownership to new system en mass which is easier than doing it disk by disk online.

MYNETAPP_ · ‎2016-05-04

without going into to much detail I believe your answer lies in 7-Mode Transition Tool 2.2, have alook.

https://library.netapp.com/ecm/ecm_download_file/ECMLP2346396

MKELLYSCSC · ‎2016-05-04

First of all, thanks for the feedback.

I was wrong about maintenance mode.

What I should have said is, interrupt the reboot process and select option 4 'Clean configuration and initialize all disks'.

My understanding is that if I do that, then, when I attach the disks to the new filer, they will show up as 'spare'.

All I need to do then is assign half of them to each node and allocate them to aggregates as needed.

aborzenkov · ‎2016-05-04

My understanding is that if I do that, then, when I attach the disks to the new filer, they will show up as 'spare'.

No; they won't show at all because they belong to different filer. And special boot menu 4 also always creates root aggregate and volume. So you will need to manually assign each disk (because if they are owned you cannot use wildcards here), destroy root aggregate, zero disks belonging to the aggregate. Exactly as if you just connected the same disks without going via special boot menu. So this step won't hurt but it is also absolutely redundant.

MYNETAPP_ · ‎2016-05-05

So you are willing to clear the disks, in that case the main thing is disk ownership. If the FAS2552 is already running C-Dot. Then simply on the FAS2240 remove ownership of all the disks, in maintenance mode. Convert the shelf to DS2246, connect to the FAS2552, take ownership, then delete the foreign aggregate if it reconginzed and simply zero spares.

Hope this helps

MKELLYSCSC · ‎2016-06-08

Thank you for the help so far.

A follow up question if I may.

The disks in the 2240-2 and 2552 are the same model (X423_HCOBE900A10).

The 2552 uses Advanced Drive Partioning to create the root aggregates.

Can I be confident that the spare drives in shelf 0 will function as spares for the drives in the new shelf?

That way I can create 2 x 12-disk raid groups in the new shelf.

Thanks.

aborzenkov · ‎2016-06-08

The 2552 uses Advanced Drive Partioning to create the root aggregates.
Can I be confident that the spare drives in shelf 0 will function as spares for the drives in the new shelf?

To my best knowledge they won't. You need to add non-partitioned spare.

MKELLYSCSC · ‎2016-06-10

Thank you for the quick reply.

Yes, you are right.

I did a bit more research and discovered that if I run the command

storage aggregate show-spare-disks

it shows that the spare disks are already partitioned.

Thanks again.

MKELLYSCSC · ‎2016-09-28

I finally got around to doing the conversion.

Before doing the conversion I removed the ownership for all the disks in the 2240.

After replacing the controller cards with IOM6 cards and cabling it all up I now have a filer with 2 disk shelves.

However, the 24 disks in shelf 1 are showing as equally distributed between the 2 nodes with all the even numbered disks in an aggregate called aggr0 on node1 and the rest in an aggregate called aggr0 on node2 (aside from the disks marked as spares).

I can see the aggregates listed when I do a 'storage disk show -shelf 1' but I can't see them when I do 'storage aggregate show'.

What is the correct way to make these disks available?

aborzenkov · ‎2016-09-28

I cannot give you NetApp link because after moving KB to new platform all bookmarked links became invalid and nothing can be found (or they cut off my access leaving it on the very rudimentary level). But here you are: http://mysysadmindiary.blogspot.de/2015/05/removing-foreign-aggregate-on-netapp.html

TomB · ‎2016-11-08

Hello,

if the disk are not zeroed before the attachment, than you will see the old 7-Mode aggregates as stale Records.

To clear the records in cDot, there is an KB Articel available.

https://kb.netapp.com/support/s/article/how-to-remove-or-access-a-foreign-aggregate-an-aggregate-made-out-of-orphaned-disks-in-clustered-data-ontap?la...

After that you should see all the disk from shelf 1 as spare. Than you can own them to any of the both controllers as you want and do an "storage disk zerospares"

This did fix this behavior on my case. 🙂

Greetings

Tom

MKELLYSCSC · ‎2016-10-06

Once again, thanks to all who helped with this particular task.

The link provided by aborzenkov has a note at the end that points out that if you are running DataONTAP >= 8.3, you need to refer to KB 1013046.

As we are running 8.3.2, I followed the procedure described in the KB.

I then had 24 new spare disks and a choice.

I could create 2 new raid groups and allocate them to the existing aggregates (or create new aggregates), or, as the existing rg0 raid groups only had 11 disks each, I could add disks to each raid group and let ONTAP partition them.

This is documented in the Advanced Drive Partitioning FAQ that I got from NetApp.

The advantage of doing this is you don't lose 4 disks to parity roles, you will lose 2 disks to additional spare disks but I can live with that.

Even with the partioning, I still end up with ~ 2TB extra disk space.

Thanks again.