NFS datastore for VM

sdrcpdpune · ‎2010-09-15

Hi,

1. Can we increase the size of NFS datastore which is in use by VMware online, without any interruption?

2. Can we move the NFS datastore to another filer with ndmpcopy or any other utility is required?

3. When I run "sysconfig -c" on our filer, I get the below output:

sysconfig: Unless directed by NetApp Global Services volumes ccvol, esxinfra, and esxdev2 should have the volume option create_ucode set to On.

Is it necessary to turn on that create_ucode? What will be the impact if they they off for some volumes?

Thanks.

martin_fisher · ‎2010-09-16

Hi,

In answer to your post:

1. In short, Yes!. Done this many a time with our VM datastore volume, which runs on NFS.

2. Do you wish to move or copy? either way you can use NDMP copy, vol copy or snapmirror (if you are licensed). Snapmirror has the advantage that the source volume can stay online whilst the transfer is in progress. Either way you would require a little downtime to switch from one volume to another, as the datastore would need to be offline to be 100% consistant. Could you not create a new volume and a new datastore and attach this to VMWare and migrate the VM's ? .. just a thought.

3. Is your appliance a cluster? leaving this setting off, could create problems in a failover or with some CIFS shares. Have a quick look at this KB:

https://now.netapp.com/Knowledgebase/solutionarea.asp?id=kb9528

Regards Martin

pborkstrom · ‎2010-09-16

Hi,

We use NFS for VMWare datastores exclusively. I wouldn't have it any other way. To answer your questions:

1. Can we increase the size of NFS datastore which is in use by VMware online, without any interruption?

a. Absolutely. You can grow and shrink the volume at will any time without any service disruption. The only gotchas I can think of are:

i. If the NFS datastore is snapmirrored to another filer, grow that volume first. After you grow the snapmirror destination, you can grow the snapmirror source NFS datastore.

ii. If you use deduplication (I HIGHLY recommend for NFS VMWare datastores), make sure you don't grow past your filer's volume size limit. FAS2020 has either 500g or 1T (depending on version). You can look it up on the NOW site.

iii. If you get an error when you try to grow the volume stating that fs_size_fixed is set, you can type the following command - vol options vmware fs_size_fixed off (run this on snapmirror source and dest), grow your volumes, then run fs_size_fixed on

2. Can we move the NFS datastore to another filer with ndmpcopy or any other utility is required?

a. If you want, ndmpcopy will work. Make sure all guests are off, but IMO, snapmirror is the best way. If you are moving the datastore to a different aggregate on the same filer, snapmirror migrate is the way to go. If it's a completely different filer, you may be able to storagemotion them to the new datastore.

3. When I run "sysconfig -c" on our filer, I get the below output:

sysconfig: Unless directed by NetApp Global Services volumes ccvol, esxinfra, and esxdev2 should have the volume option create_ucode set to On.

Is it necessary to turn on that create_ucode? What will be the impact if they they off for some volumes?

a. The sysconfig -c error - I saw that on my filer. You can change that option on the volumes HOT, anytime. I'm not positive of the impact of it being off, but the data is "safer" with it on. I called netApp on it when I first got into filers and noticed it. I changed it in production during peak hours and had no problems whatsoever.

-pb