This KB pertains to LAN free, Fiber/iSCSI based, restores only.
LAN free backups, over fiber/iSCSI, as well as file level restores are successful.
Full restores look to be successful but renders VM with no OS or any data at all on disk.
Full restores over the network are successful.
This is generally caused by LUNs presented to the vRanger machine as read only and/or are offline.
The vRanger deployment guide advises that vmfs LUNs be presented to the vRanger proxy as read only when applicable for backups. Not all NAS devices offer this option.
The guide further explains that when doing LAN-free fiber/iscsi based restores that the LUNs must be presented as read/write:
• On your storage device, zone your LUNs so that the vRanger HBA has Read-Only access (for backups). If you want to restore over fibre/iscsi, zone your LUNS so that the vRanger HBA has Read-Write access.
To get around this issue.
To check from Diskpart on 2003 machines; open a command prompt and type 'diskpart'. From the Diskpart prompt type 'list disk'.
From 2008 machines from diskpart prompt type 'SAN'
If any of the disks are showing as offline bring them online from disk management or from diskpart.
From diskpart prompt on 2003 machines type 'online disk #'
On 2008 machine from diskpart prompt type command 'SAN POLICY=OnlineAll'
As presenting all of your vmfs LUNs to the vRanger box as read/write can be risky, we suggest the following best practices:
1. Leave LUNs in read-only mode until the need for restore. If a restore is needed change to read/write
2. Create a seperate vmfs LUN presented to a single ESX host and the vRanger proxy present as read/write and restore there. Once restore is finished vmotion restored VM to intended target. Or at time of restore prsent LUN to target host. When doing this for the first time on a newley created vmfs volume you might have to first create a dummy VM (no OS) to get LUN to initalize.
Full vRanger deployment for current 5.3.1 GA release:
VMWare SAN transport best practices:
Windows 2008 san policy online ref:
The reason why the jobs look to run successful instead of failing when we cannot write is under investigation.