Hi, Im pretty new to the world of booting from iSCSI Luns and wondered if I could ask some basic questions in hope of some guidance or examples of how someone else would do it?
What Im trying to achieve is to have a Windows 2008 server booting from a LUN and once running have it access several more “data” windows drives via iSCSI using the MS iSCSI initiator. Pretty simple I hope.
The service profile has currently 2 vNICs, one for iSCSI booting and one for standard network access on the production network. The iSCSI boot vnic has just the iSCSI VLAN and the other just the production network VLAN.
I’ve managed to work out and build a 2008 R2 server on a B200 M3 blade booting successfully from a single vNIC and overlay iSCSI vNIC and have it up and running so that part is done and understood.
What Im wondering is, the iSCSI storage is on a single VLAN so I don’t see a point in having 2 x iSCSI boot vNICs and the single one is using failover in event of a fabric outage. Im not 100% if the storage IP and IQN remain available in even of a NetApp storage controller failure but that is something Im looking into with regards to protecting the server against a storage failure.
Does anyone else run with just the single vNIC for booting and with regards to accessing further iSCSI LUNs to map as Windows drives, is it recommended to have a third vNIC for this iSCSI traffic, i.e. not to use the iSCSI vNic used for booting or does it not matter?
Another issue Im having is that within the OS, I see the IP address Ive used for the iSCSI boot in the service profile is shown in the OS under a Ipconfig /all as being there for one of the adapters (ive not used a DGW or other details as it is on the same iSCSI VLAN as the storage) and Ive given the other windows adapter a free IP address from the production network. Ive made sure there is only the one DGW used in the OS but cannot ping the server from outside the VLAN or ping from the server out. Has anyone else ran into this sort of issue?
Sorry this description is a bit labored but thank you in advance for any advice.
You can use the boot iSCSI HBA to get your data LUNs. Unless you need to force traffic to come from different paths or for performance reasons just use the same boot HBA.
If you're running 7-Mode OnTAP you will end up having the IP address targets fail from one controller to the other in the event of failure (ALUA is not supported for iSCSI in 7-Mode apparently). So your IQNs and IPs will not change for a controller failover.
As for the IP address, I assume your iSCSI VLAN is different a different, unique subnet from your "production" network traffic? Windows Firewall up? How did you configure the VLAN configuration for the vHBA? Is it sending a native VLAN down to the OS or is it just a checked VLAN and native not selected?
Thanks Steven, the network access was down to not having the VLAN tagged as native so Windows was ignoring the packets, all sorted now!
I have both vNICs showing in the OS. The production network adapter/vNIC and the vNIC used for iSCSI booting. An IPCONFIG shows the IP used in UCSM for the iSCSI boot so I re-affirmed that in Windows by assigning the same IP address and assigned a different IP address used to communicate with the server to the production adapter. I assume once I configure the iSCSI initiator and start adding LUNs to map as Windows drive that all iSCSI traffic will go via the iSCSI vNIC / adapter as this is the only one configured with the iSCSI VLAN?
Also, as I have only 1 iSCSI vNIC, do I still need to install and configure MPIO?
iSCSI vNIC only exists during the boot up phase and is not available to OS after the boot.
All other iSCSI data traffic is carried over NICs provisioned with iSCSI VLAN after the boot.
UCSM version ? Is it VIC adapter ?