Subscribe
Accepted Solution

help breaking up a giant volume

Hi,

I've got a 3140 filer, running 7.3.4. I have a giant volume(7.5TB), no qtrees, it's one giant mother of a thing that's in production use(NFS). I wondering if there is anyway to break this thing up into manageable smaller volumes or qtrees while minimizing the outage window to do so.

I have another 3140 as well, both have flexclone, snapmirror and Multistore licenses and are connected by 10GBit ethernet.

Re: help breaking up a giant volume

I've been giving this one a bit of thought since you made your post.

I'll hope the data is at least segmented by directories ?

First, move the data in the volume from directories into qtrees.

Create the qtrees

Use 'mv' from an nfs host to move files into the qtrees.

(Usually moving files within the same file system is not the same as copying and then removing the original. First a new  link is added to the new directory then the original link is deleted.  The data of file is not accessed. This is much faster than copy and  remove. The file still has the same inode.)

Change your exports and mounts for the data to the qtrees.

(you might try using the acual= export option to avoid changing client mounts)

Now that the data is in qtrees you can use qtree snapmirror to copy the partitioned data into new volumes with qtrees, and synchronize with snapmirror.

Change your exports and mounts for the data to the new qtrees in new volumes.

Delete the data in your source qtrees.

'vol size' to shrink your 'giant' volume back down in size.

This is the best option I could think of to minimize downtime.

I hope this response has been helpful to you.

At your service,

Eugene Kashpureff

NetAppU Instructor and Independent Consultant
(P.S. I appreciate points for helpful or correct answers.)

Re: help breaking up a giant volume

I'm going to do a test run to see how long that would actually take.

Another option I've uncovered is this one:

Use FlexClone to clone the volume. However many volumes you want the data separated into at the end, clone it this many times. Do it with no space guarantee on the volume. Now, create a NFS share for yourself to each of the volumes. Delete all other data that exists in the volume. Deleting data will be quicker than copying as you don't need to read, interrogate and re-write all the ACL's and file trees, just delete . Once this has finished, break the clones (will use less storage now), fix the space guarantee and repoint the NFS shares.

This one looks very elegant, however I'm wondering whether to break the clone link before or after deleting files in each clone? Also, with 85% ofthe underlying aggregate used, can I be sure that this operation with flexclone won't take up any more space?

Ivan

Re: help breaking up a giant volume

Clones would be a cool solution also.

You'd want to split the clones AFTER deleting files.

You'd need to mount the new volumes directly after cloning, to avoid the need to resync any changes to data...

I hope this response has been helpful to you.

At your service,


Eugene E. Kashpureff

Fastlane NetApp Instructor and Independent Consultant
(P.S. I appreciate points for helpful or correct answers.)

Re: help breaking up a giant volume

One more question, will I be able to turn on SIS once I've deleted the files and shrunk the cloned volumes? Currently I can't since it exceeds the size limitations.

Thanks for your help, points awarded.

Re: help breaking up a giant volume

Yes, you should be able to dedupe your volumes when they're small enough.

You'll be going volume by volume:

Snapshot the parent volume.

Clone a volume off from the snapshot.

Re-mount the section of the cloned volume you want.

( You may want to create a qtree and still move data into a qtree on the volume. )

Delete the directories you don't want.

'vol clone split' that clone off the parent.

Delete the section you've cloned off and split from the parent.

Delete the snapshot you cloned from.

Run sis on the cloned volume.

Shrink the parent volume.

Repeat for each directory you want to split off...

Run sis on the parent when it's small enough.

For extra measure run 'reallocate measure' on your volumes and reallocate as needed.

Thank you for the points !
I hope this response has also been helpful to you.

At your service,


Eugene E. Kashpureff

Fastlane NetApp Instructor and Independent Consultant