With ESX 4.0 vmware has made changes pertaining to the functionality of the LVM.DisallowSnapshotLun and LVM.EnableResignature settings that were available in 3.5 under the Advanced Settings. With ESX 4.0 you are asked if you want to allow these settings when adding a LUN that is a snapshot. You can however edit these settings at the host level in the esx.conf file to make this a "global" setting at the host level. This can be done in ESXi as well using "unsupported" mode.
Our issue is that we utilize NetApp Snapmirror and Flexclone technology to provision test LUN's to our DR Vmware environment for testing purposes. We have the entire process scripted to provide failover to the DR site for both a testing environment (using Flexclones) and for an actual "takeover" on the DR SAN in the event of an actualy disaster. Think of it as SRM without the GUI only the failback process is a more streamlined process.
With these new settings we've discovered several things.
A. Simply changing the LVM.DisallowSnapShotLun = 0 and rescanning the ESX host does not "mount" then snapshot LUNs. The storage adapter recognizes the LUN but the Host does not mount the LUN. However, rebooting the ESX host will mount the LUN properly.
B. Another observation we found is that even using the GUI to apply these setting did not work. It returned an error "Cannot configure Host Setting".
We have opened a case with Vmware and they have observed these behaviors and they do not have a resolution as of yet.
My question is first of all has anyone else ran across this issue and secondly, is there a way to flush the memory on the ESX host by restarting a service? Vmware has stated that this information is being stored in the host memory yet running /sbin/services.sh restart does not resolve the issue. We have to reboot the host everytime we fail these LUN's over and then fail them back. Even after performing a rescan.
Any information would be helpful.
