Creating GFS2 data store fails on VME 8.0.4 / Ubuntu 22.04

[Aqid]

Hi all,

I currently have 2 HPE DL360's and a single HPE Nimble HF40. One of the DL360's is running vSphere 8.0.3 with vCenter 8.0.3, and the other is running Ubuntu 22.04 with VME 8.0.4.

I created 2 LUNs on the HF40 and presented one LUN to both hosts, and one LUN to Ubuntu only. The one that is presented to both hosts is formatted as VMFS.

I went through the iSCSI and multipath setup in Ubunu. I connected/logged in to both LUNs. One is the VMFS LUN and is shared with the vSphere host, the other is RAW and had never been connected to. When I run iscsiadm -m session I see all my multipath sessions (4), and when I run lsblk I see 8 partitions as "mpath" with 2 unique mapper IDs.

When I go into VME, the first LUN with the VMFS datastore mounts just fine and it shows up under the vCenter cluster.

As next step, I went to Inventory -> Cluster -> Selected my VME Cluster -> Storage. And hit Add storage and select GFS2. Once I select that, the correct mapper ID pulls up. 

iscsiadmin session and lsblk outputcorrect mapper ID showing when trying to add

Everything looks good from here, but when I hit Save, nothing happens. It just gets stuck on this page, as if I haven't clicked the button..

I inspected the logs on Morpheus and this is the only related thing I could find:

log output

I even tried creating the GFS2 partition using the command line, but as this was not in HPE's support documentation I don't know if the command is correct. The command executed but the LUN did not show in VME, so I think I will have to undo that step again.

That's about as far as I got, any insights would be MUCH appreciated! Thanks.

[DiegoDelgado]

Hello,

if I understand correctly, you are trying to add a GFS2 datastore to a single node VM Essentials cluster, am I right?

If this is the case, you're not able to finish the task because we need 3 nodes for a shared GFS2 datastore, I have not tried with a single node, but if you only have 2, an error message shows stating that you need a 3rd node to be able to proceed. Can you please confirm if this is your case so we can pass your feedback so engineering can add the warning in this case too?

[Aqid]

Thanks for your reply. You are correct, I'm trying to add the data store to a single node (that will be a 2 node cluster later). I'm confused on why I can't add an iSCSI LUN to a 1 or 2 node cluster? On ESX I can create a VMFS datastore on an iSCSI LUN directly on a single node, on Windows I can create an NTFS or CSVFS volume on an iSCSI LUN on a single node. Why wouldn't I be able to do that with Linux/GFS? If this is truly the case, it immediately disqualifies the product for our entire company. We exclusively use 2 node clusters with Alletra back-ends accross our entire install-base., and our idea of migrating them from VMWare to VME was to reinstall one of the nodes with VME and have the other one on VMWare during the migration, hence my lab setup.

Given this limitation, can you think of any other scenario we can use VME with a 2 node cluster with an Alletra SAN? And can you think of any scenario where you can migrate from a 2 node + Alletra VMWare cluster to the same 2 node + Alletra cluster on VME?