#ADPv2 move/copy to smaller drives?

1 messages · Page 1 of 1 (latest)

unborn falcon
#

ADPv2 quesiton:
We have AFF400 version 9.12.1P2.
4 shelves (ds224C) of x319 drives (7325.4GB) and
9 shelves (ds224C) of x357 drives (3662.5GB).
During initial config the large drives were used for ADP root-data-data.
So there are 48 drives currently used for ADP of the x319drives.

  • We want ADP on the x357 drives. (FYI we duplicate netapp HQ cluster aff400 that was configured using the x357 drives for ADP)
    Is there a way to move the ADP to the smaller drives without destroying it and reconfiguring? (System already in production) I can free up 48 spares of the
    smaller drives if it is possible.

Reason:
We need to make sure that we end up with both 3.8 (whole and partitioned) as well as 7.6 (whole) spare drives on both controllers
when all is said and done (like there is on HQ controller). Otherwise, we’ll end up with a low spares condition and it’ll pull in the wrong
size disk and it is a pain to fix.

lyric meadow
#

@tidal ibex Found you an ADP question!

unborn falcon
#

Thx Nick

tidal ibex
#

The road is long, windy, and difficult but you can get there.

#

The easiest way to get there, though, is to reinit the controllers. It's not ideal, but it is the easiest to accomplish this.

#

First: Run 9a against both controllers. This will vaporize the configuration.

#

Second: Disconnect all shelves with 7.6TB drives in them. Only the shelves with 3.8TB drives will remain connected.

#

Third: Run option 9a again on both controllers. Then, option 9b. (9a is a pre-req for running 9b and cannot be skipped over.)

#

This will implement ADP on 48x of the 3.8TB drives. Once the cluster is back online, then re-attach the 7.6TB shelves. They will remain whole and auto-assign themselves as expected.

#

If that's not ideal, and I understand that it's already in production, you're going to want to open a support case. There are too many moving parts for a text-based chat.

unborn falcon
#

Yep, support case has already been opened but NO answer yet.

#

Starting over isn't an option unfortunately. I have reinit many controllers and wish this would have been caught earlier. So as you see it there isn't a way to do this hot? Only way is to wipe it out and restart?

lyric meadow
#

You’re completely rebuilding disk layouts. Doesn’t matter if it’s ADP or not, you don’t want live data anywhere near that, no matter the vendor or platform

#

Do you have the excess capacity elsewhere to middle man it while you rebuild the layout?

unborn falcon
#

Correct. That was where the spares come in. If they were the same size this wouldn' t be an issue.

#

we have excess capacity but sounds like I have to evac the whole thing and wipe

#

Thanks for responding. I apprciate the help from both. I will update you once I hear back from Netapp. Glad I found Discord.

lyric meadow
hollow laurel
#

I just sorta did this yesterday. Not sure who all is able to access this kb, @unborn falcon you may want to ask your support person to refer to:

https://kb.netapp.com/onprem%2Fontap%2Fhardware%2FHow_to_partition_a_drive_manually_in_ONTAP_9

Even though you are running ONTAP later than 9.5, I used those instructions (the ones labeled ONTAP 9.5 and earlier) on 9.10.1P12 and it worked.

Honestly, it’s not something I would want most customers to do without a little handholding.

NetApp Knowledge Base
How to partition a drive manually in ONTAP 9

Partition a drive manually in ONTAP 9 – If a disk is not partitioned automatically or is partitioned wrongly, manual partitioning is required. The target disk must be of the same size and type …

#

I also wish Netapp would update the cluster shell command to allow easier creation of partitions (within limits)

tidal ibex
#

@hollow laurel And this is exactly the "long, winding, and difficult road" that I mentioned earlier. Yes, it is technically possible. It's far from simple, as you have pointed out.

unborn falcon
#

Reading that doc says target disk must be same size and type. My goal is to move all ADP drives from 7.6tb "down" to 3.8tb without a down time. I have spare 3.8 drives. Is it possible?

analog forum
#

If you dont have any disk with the same size and the desired partition layout, you need to partition it manually. And with manually I mean manually manually. In the nodeshell with diag privilege. That's why Scott said it's not that easy if you don't know what you're doing.

#

If the system is under support, create a case and ask for help.

#

Also I'm not sure if the automatic root-migration cmd (system node migrate-root) will work correctly with partitioned disks. In my experience this cmd will do like 90% of the way correctly but then often fails at some point... which means you need to know how to fix the rest.
Otherwise you would need to do the whole root-migration to new disks manually too. Which involves many and quite some heavy cmds.

#

Can you explain why this is really necessary? I dont see any isseu with your current ADP design. The root aggrs don't really care if they are on 3.8TB or 7.6TB disks. They will be the same size anyway.

unborn falcon
#

HI OG1, I am PS Eng with Partner. 2 clusters ordered for customer and one cluster came with 7.6 ADP and the other 3.8 ADP. Need them the same.

#

Didn't catch it until in prod.

analog forum
#

Use the part under "ONTAP 9.5 and earlier". You can't use storage disk create-partition because you don't have any already partitioned disk with the same size.

#

Wish you good luck

unborn falcon
#

THANK YOU!!!!!!!!!!!!!!!!!!!!!!!!!!