Hi,
I have a cluster cosisting of five hosts (host01 - host05), with 10 LUNs presented to the hosts, and all hosts see the LUNs in the same order, i.e., each LUN is seen as the same vmhba # and same LUN ID # across all hosts.
I've created a new cluster with one host (host06) so far. I zoned and formatted a new LUN to this host, and it shows up as LUN ID 0 and vmhba2:0:0 on that host. I need this new LUN to also be accessible by the hosts in the other cluster. But here's the problem - after zoning the fabric, when I rescan HBA on hosts01-05 they see the new LUN, but when I refresh storage, they're not seeing the VMFS volume.
A VMWare engineer told me that the problem is caused by the new LUN being assigned a LUN ID of 0 by host06, and when hosts01-05 refresh storage, they're not allowing the new LUN to come online because their next LUN ID would be 10, but host06 has assigned it LUN ID 0, so they think it's a snapshot volume. As a test, when I set the LVM DisallowSnapshotLun setting to 0, hosts01-05 can see the new volume, but I'd rather not keep that setting at 0 (I've changed it back to 1). The engineer advised me to either contact my storage vendor to see about manually changing the LUN ID to make it same across all hosts, or to zone the ten existing LUNs to the new host in the new cluster so that it'll have the same order of LUNs.
I have some questions about this. Is it the vmhbaX:Y:Z or the LUN ID (0, 1, 2, ....) that must match across hosts? Is there a way to change this on ESX, instead of making a change on the SAN or doing the work of zoning the other ten LUNs to the new cluster/host, which I don't need? Moving forward, what if I wanted to map a LUN to a subset of hosts rather than all of them? Wouldn't that cause the same problem of mismatching LUN IDs? I've searched through the VMTN forums, and while I've found some really good information (as usual), I haven't found a thread that explicitly answers my questions.
Thanks in advance for any insight!