How to move a volume containing qtrees and figure out the downtime needed?

[ACHOU_SIMG]

Hi,

OnTAP 8.03P4, 7-mode

I have a volume I need to move to another aggregate so I can retire the hardware used by the volume. I think I can do one of the following:

option A: do 'vol move' for the volume

  question 1: I repeatedly see articles saying 'vol move' is non-disruptive but the volume must not be exported during the 'vol move' operation. Isn't it disruptive from the users' perspective since they can't access the volume?

  question 2: Isn't the downtime from users perspective the time it takes to do 'vol move' which will take hours for a mutil-TB volume?

option B: move qtrees in the volume one by one to another volume

  question: I learned I should do qtree snapmirror as part of qtree move. However, Once I've done QSM and bring destination up to date, how do I "switch" to the qtree to the new destination?

Thank you.

[JGPSHNTAP]

You mean you are moving to another controller?   What does need to move volume to another aggregate to retire hardware.  I assume another controller.

Please clarify then we can help

[ACHOU_SIMG]

Hi,

Thanks for the reply. The volume will be moved to another aggregate hosted by the same controller. I have many Linux clients connecting to the filer via NFS. I was hoping to use 'vol move' which does not change volume name thus no change on Linux clients required. However, from what I understand 'vol move' requires NFS volume unexported thus inaccessible during the move operation and the volume is several TB so it seems to me from the Linux client's perspective this means it will take hours of downtime during the move.

The end goal is to remove the shelves containing the volume after the volume is moved.

Thank you.

[JGPSHNTAP]

How do you plan on taking the old hardware out of service?  You can't simply disconnect shelves.

Vol move is better for LUNS

If you need to migrate the data to new volumes,  what's wrong with snapmirror to a new volume

[ACHOU_SIMG]

Hi,

If I snapmirror on the same controller, I have to use a different volume name which means I have to change automount maps on the clients. Somehow in my case, some Linux client's automount does not always pick up the changes (could be because the volume was used I think).

Please correct me if I am wrong.

I will take out the shelves when the filer can be brought down for maintenance. I guess you were suggesting using snapmirror and when it's time to bring down the filer, do the final sync.

Thanks!

[JGPSHNTAP · Accepted solution]

Yes