REALLOCATE volumes after grow aggregate in CDOT

[heightsnj]

I know that we should run reallocate after more disks are added, I also know that we should run it in volumes basis not on aggr level.

Here is my question. We have thousands of volumes, if we run the following command, then we wil have to keep track of all volumes and manually run through the list one by one. Yet, in SM situation, we will have to break SM relationship first, and go through the same in DR site.

It sounds very time consuming. will this be the right process that we will have to follow?

reallocate start -f -p –vserver vserver-name –volume /vol/volume_name

[heightsnj]

Any inputs please

[ekashpureff]

Heightsnj -

My best advice is not to grow aggregates.

Plan your aggregate layout and do it right to start with.

Or if you have to, do it with lots of disks to avoid creating hotspots.

ONTAP 8.X allocates the storage space in the aggr a bit different than 7.X.

Hotspots from adding disks are not as much of a problem as it was in earlier versions of ONTAP.

There is the undocumented hotspots= output from the reallocate command to see if you do have hotspots on a volume after adding disks.

If you have thousands of volumes and you wanted to measure/reallocate on all of them, then I'd automate the process with the APIs.

I don't have a sim up and running right now, and the reallocate command is conspicuously missing from the manual page refernce for Cluster Mode.

(Though it does show up in the 7 Mode to Cluster mode command map?)

Can't verify your syntax for reallocate start.

I hope this response has been helpful to you.

At your service,

Eugene E. Kashpureff, Sr.
Independent NetApp Consultant http://www.linkedin.com/in/eugenekashpureff
Senior NetApp Instructor, IT Learning Solutions http://sg.itls.asia/netapp
(P.S. I appreciate 'kudos' on any helpful posts.)

[heightsnj]

Hello there, Thanks for your inputs.

We have just added 2 new shelves, in addtion to 4 already configured ones which made up one large aggr. So, my question is, should I add these two shelves/disks into the existing aggr, or should I use these new two shelves to create a separate aggr.

I heard that one large aggr should come out better performance, in this sense, I should add two new ones into the existing aggr, thsu, we will only have one large aggr. The issue with this is that we will have to run "reallocate" on all existing volumes.

The other way is to create seprate aggr which includes 2 new shelves only. Thsu, we will not run "reallocate".

Please let me know your thoughts.

Thanks!

[snthlkumr]

Check this KB

https://kb.netapp.com/support/index?page=content&id=1014680&locale=en_US

[heightsnj]

Thanks to snthlkumr , this is very helpful in saving efforts by scheduling Reallocation jobs.

However, since this is production environment, and sensitive to I/O activities. Plus, we have quite a few SVM's and each has quite a lot volumes, even with scheduling Reallocations, still it will be a long process.

 I am thinking if it is worth it to add two new shelves into the existing aggr, or just to create a separate aggr then to avoid run reallocations. How much difference between these two method?

[ekashpureff]

Heightsnj -

Yes, large aggrs with lots of disks are a good thing.

But 80+ disks in your aggr already is large, and so would be using two shelves to create a new 40+ disk aggr.

I'd create the new aggr if I was in your place.

I hope this response has been helpful to you.

At your service,

Eugene E. Kashpureff, Sr.
Independent NetApp Consultant http://www.linkedin.com/in/eugenekashpureff
Senior NetApp Instructor, IT Learning Solutions http://sg.itls.asia/netapp
(P.S. I appreciate 'kudos' on any helpful posts.)