Hi there!
As stated in KB 1005930 (and several other places as this communities), the UUID of a NFS datastore is generated from a hash between server and folder. In my case I have all my VMs in some NFS datastore mounted by IP. Now I need to have everything mounted by FQDN, and for this purpose I have modified the host file on the ESXi. Once I mount NFS with FQDN, it is named with a (1) at the end, as said.
Now, my plan is to Storage vMotion everything. I have an empty ESXi with the NFS mounted by FQDN, and so named NFS01 (1). It is the same vol on the nas than NFS, but in VMware they have different UUID. My fear is if this Storage vMotion can suppouse any locking or corruption, as it is moving from one datastore to itself. I want to try this way because it's preferable avoiding cold migration (and stopping all VMs).
Working on 5.1 build 799733.
Any tip?
Thank you very much in advance for your time ![]()
David
By the way, I understand that the best option is to mount a completly new and different NFS datastore, migrate everything there, unmount&mount NFS datastores, and migrate back. But not enough free space for that ![]()
I apologize for the double-posting, but could find the "edit" button
How many NFS datastoes do you have? How big are your NFS datastores? and How many VMs do you have per datastore? I ask because the other way to approach this with your constraints is shut down VM on datastore, unregister the VMs from inventory, unmount the datastore, and remount with fqdn and reregister the VMs again and power-on your VMs and you should be in business. But of course this requires some downtime.
Yes, the costumer is finally moving into that idea. Shutting down everything, unregistering, unmount&mount, register and power on. I was just trying to avoid the downtime.
Five datastores, huge ones: 3Tb for the smaller, 15Tb for the bigger. Lot's of VMs inside each of them.
Thanks a lot! ![]()
