We've been using iscsi vmfs luns on our netapp fas270c to provide disks to our vmware servers. We've recently upgraded our fas270c to include Fibre Channel and would like to change the luns to use FCP instead of iscsi. On the Netapp side, we've mapped the LUNS to use both an iscsi id and an FCP id.
The problem I now have is getting our ESX3 servers to use the FCP LUN id instead of the iscsi. The ESX servers can see the FCP LUN id's but if I try to asign them to a disk, I get a warning that the disk being created will be wiped.
I'm unable to find any documenetation on how this should be done. If anyone's done this already and can either give me instructions on what you do or point me in the right direction I would very much appreciate it.
Thanks
Niels
OK,
Did you have the same lun ID for your iSCSI LUN and your FCP LUN?
Try to make a refresh on Datastore tab.
Hi,
if i understood your problem, you can see your LUN on you ESX fiber card and you cannot see the Datastore created before.
Try to enble vmfs resignature in advandce setting of your esx server.
GO to:
VCS -> ESX server -> Advenced settings -> LVM -> LVM.EnableResignature-> set this option to 1
-> scan you FC card and make a refresh for your datastore tab. Normally we will se your datastore with a new name as Snapshot of ...... . Rename the datastore.
You can change LVM.EnableResignature to 0 when Datastore is visible.
I hope that this information could help you.
Thanks.
Almost there...
Thanks for your help so far but I'm still not quite done...
After changing LVM.EnableResignature to 1 and refreshing my FC HBA, as you expected the iscsi datastore was renamed to snap-00000007-fas270c-1_vm_testsystem.
However, the FCP datastore which also maps to the same lun on the Netapp box is not in the list of storage devices. so I can't rename them to what the datastore should be.
As mentioned previously, if I try to use the Add storage option, it tells me that the current disk layout will be destroyed.
Any more suggestions?
Cheers
Niels
One step closer still.
I removed the iscsi mapping on the Netapp box for the lun. This then reflects the correct datastore ID on the esx server. However when I try to rename the datastore to it's previous setting ESX server tells me that the datastore name is already in use. The VM that the datastore is used by is greyed out and says inaccessible on it.
Any ideas?
Cheers
Niels
OK,
Did you have the same lun ID for your iSCSI LUN and your FCP LUN?
Try to make a refresh on Datastore tab.
Thanks pcomo, your help has been very appreciated and I'm starting to move
our servers over as I type
Just to outline our layout and what we did to finish it for anyone else that
needs to do this...
We have two ESX Servers that both connect to our Netapp fas270c via iSCSI
for the VM's storage. The servers are configured to failover to each
other should one of them go down. We recently upgraded to FCP and we
needed to move our LUNS over from iSCSI to FCP.
1. On the ESX Server Virtual Infrastructure Client you need to go to the ESX
Server hosting the VM.
2. Click on the configuration tab, select 'Advanced Settings' from the
'Software' pane.
3. In the window that appears, select LVM and under the setting
LVM.EnableResignature change it from 0 to 1, click on OK to close the window.
4. Now, on the Netapp Filer, we removed the iSCSI LUN ID Mapping and added a
mapping to the FCP using the same LUN ID that was being used by the iSCSI LUN (This
allows the change to happen almost seamlessly)
5. Back on the ESX Server we went to Configuration>>Storage Adapters
and rescanned all HBA's and VMFS Volumes. After the first rescan you will
notice that the iSCSI LUN is no longer there but the FCP LUN has not been found
either. Also, the VM will be marked as 'inaccessible', if you check the
Storage you will notice that the volume that used to be assigned to the VM now
begins with 'snap...'
6. Rescan your HBA's and VMFS Volumes again, you will notice that the FCP
HBA now picks up the LUN and your VM will once again become accessible.
7. Restart your VM and away you go.
8. Don't forget to change the LV.EnableResignature back to 0 once done.
If you don't/can't use the same LUN ID for the FCP HBA then you are going to
come across the situation where the LUN doesn't get reassigned to the VM.
To counter this I followed the above steps but after point 6 you need to
also do the following.
(By the way I hate doing this as I've had a couple of hairy moments where I
thought I'd lost the VM... I didn't but I wont be held accountable if you
do...).
7. Right click the inaccessible VM and select remove from inventory.
8. Under the ESX's server Configuration Tab, Under Storage, find the Datastore
that used to be used by the VM. It will begin with 'snap...' Rename
the Datastore to whatever you want to call it, I suggest setting it to its
original name. Do this to any additional Datastores that may have also
connected to the old VM. (If you get a message telling you the datastore name
is already in use, remove it from the list of datastores using the 'Datastore'
option from the dropdown on the views button.
9. Right click the renamed Datastore that stored the OS and select 'Browse
Datastore'.
10. Browse the datastore and find the location where the OS resides, it's
obvious on our configs, your may be different.
11. Find the file in that folder ending with '.vmx', then right click and
select add to inventory. A wizard will pop up asking you for the name of
the VM etc. Follow the wizard and your VM will reattach itself to the
Datastore.
12. If the VM had multiple HDD's all connecting to various Datastores,
you'll need to go into the VM's settings, make a note of the Virtual Device
Node for each unattached HDD, remove the unattached HDD's and add new HDD's
connecting to existing datastores and attaching them to the correct Virtual
Device Node for each HDD.
13. Click on OK and the HDD's will re-attach themselves to the Datastores.
Done.