combine or move disks from a raid group

[chris_lawless]

Hello,

We created a raidgroup size for an aggr of 5 disks instead of leaving it the default 16.   We now have 2 raid groups, rg0 - 5 disks, rg1 - 4 disks.  I know we can expand the raid group size, however is there a way to eliminate rg1 and move, replace, or recover its disks with out blowing the whole aggr away?

FAS3270 - ontap 7.3.6

thanks!

--Clawless

[billshaffer · Accepted solution]

No, once the drives are part of an aggregate, they can't be taken out (with the exception of failing them, of course - but that won't accomplish what you want).  I have wanted to do similar things many times - it'd be nice if you could balance drives out like you can with Oracle ASM disks...

Out of curiousity, why did you choose a RG size of 5?

Bill

[chris_lawless]

Thanks Bill - thats what I was afraid of.

It ended up 5 by accident / lack of attention to detail I think - this filer is at our DR site and it’s also an office, our support guy there has been learning the NAS as a backup, we were moving all of his share data to the NAS from a file server and I told him to create an aggr with 5 disks, which he did and has done so before, the issue just came up when we needed more space, I went to add another disk and noticed the raid group issue.  I guess he changed the raid group to 5 thinking it was the number of disks, and he apparently added three disks on his last expansion.  I went back and checked all the other he has done and they are correct – not paying attention I guess.

Yes a way to reallocate disks / RGs or shrink aggrs would be nice. I have several times needed to shrink / recover disks from an aggr after a project is over but there is just a little bit of data remaining, so I have a catch-all aggr that I mirror the last bits to until they can be deleted so I can blow away the original aggr.  Not hard just time consuming - easy to give, hard to take back.

[JGPSHNTAP]

If this is a DR copy of your data, and you have extra disks, create a new aggr and snapmirror the data over

[chris_lawless]

Good plan!  We are setting that up now.