FlexGroup "volume expand" question | NetApp | Page 1

low bluff Jul 19, 2024, 2:29 PM

#

Adding new aggregates to a full flexgroup vol

"volume expand.. -aggr-list" Does the aggr-list need to include the existing aggrs, or just the ones we're adding? Docs unclear.

knotty sparrow Jul 19, 2024, 2:38 PM

#

hm. the man page says this:

-aggr-list <aggregate name>, ... - List of Aggregates for FlexGroup Constituents

Specifies an array of names of aggregates to be used for new FlexGroup constituents. Each entry in the list will create a constituent on the specified aggregate. An aggregate may be specified multiple times to have multiple constituents created on it.

I think it's pretty clear in that you only need to specify the new aggregates. Which document did you refer to?

low bluff Jul 19, 2024, 2:48 PM

#

ah, so if i want 12 new vols on aggr1 and 12 on aggr2, -aggr-list aggr1,aggr2, x12 times

glad stone Jul 19, 2024, 2:50 PM

#

volume expand -volume abc -aggr-list aggr5,aggr6 -aggr-list-multiplier 6

#

it should also tell you what it will do before you commit and prompt for a confirmations (unless you turned off confirmations!)

low bluff Jul 19, 2024, 3:07 PM

#

got it!

New issue is this vol has 24 constuents currently, 20 are ~5.2TB, and the last 4 are 2.8TB for some reason. vol expand wants to create new 2.8T consituents. I want 5.2TB.

#

(and thanks for the multiplier tip, actually copy pasta'd the list 12 times. LOL)

low bluff Jul 19, 2024, 3:42 PM

#

looks like I need to " vol modify -vol fg_vol -max-constituent-size 5.2TB"

low bluff Jul 19, 2024, 4:15 PM

#

hmm, that's unrelated, vol -expand just picks smallest existing. Ah well, just need to have more vols.

knotty sparrow Jul 19, 2024, 5:01 PM

#

I really miss the simplicity of good old FlexVol HPO.... dunno why that was not carried over from 8.x

#

this whole FG elastic sizing, FG rebalancing, extend vs. expand and stuff feels like it was bolted on with lots of duct tape 😦

low bluff Jul 19, 2024, 5:07 PM

#

I feel that way about anything that isn't FC or NVMeoF

knotty sparrow Jul 19, 2024, 5:10 PM

#

yeah, well, FC is not the magical panacea people make it seem to be. For example, many Brocade FabOS versions had issues with big MetroCluster FC backends where the sheer number of devices (every disk logged into the fabric) would overwhelm the FabricNameServer

#

fun times 🙂

low bluff Jul 19, 2024, 5:12 PM

#

yeah, using TOR switches to manage hundreds of devices, lol

glad stone Jul 19, 2024, 6:02 PM

#

you might be able to do this after the fact...

#

expand the flexgroup, then see what size it is. Figure out roughly how much more you want to add (looks like roughly 28.8T) and then do "vol size flexgroup +30T". I suspect it will resize all constituents to be about the same. Jut a guess/hunch though

low bluff Jul 19, 2024, 6:21 PM

#

Created 36 new constituent vols, for ~100TB. Can't do the vol size thing, bc the pre-existing aggregates are nearly full.

glad stone Jul 19, 2024, 6:43 PM

#

Did you try? I think ONTAP may take that into consideration

low bluff Jul 19, 2024, 7:47 PM

#

No, it's a production system serving an essential manufacturing process, I wasn't going to just try things. LOL

scarlet maple Jul 22, 2024, 11:39 AM

#

why didn't you just move some of the constituent volumes to the new aggregate? Then you could have expanded without creating new volumes since they seem to be quite small

low bluff Jul 22, 2024, 1:57 PM

#

time. Would have taken several days, but even with time spent clearing up my confusion about the commands, the flexgroup was expanded in a couple hours.

The existing vols were %99 full, production lines were at risk. Was an emergency reallocation of disks/shelves to address.

#FlexGroup "volume expand" question