SOLVED
Assuming the ioscan output was from within the VM guest itself, that is expected output. NPIV allows storage to be "presented" directly to the VM guest but is not the same as SR-IOV or Direct I/O where the FC port itself is dedicated to the VM. The port is still virtualized inside the VM guest and associated with the GVSD driver:
# ioscan -fnk
Class I H/W Path Driver S/W State H/W Type Description
============================================================================
ext_bus 6 0/0/0/0/0 gvsd CLAIMED INTERFACE HPVM NPIV Stor Adapter
/dev/gvsd6
ext_bus 0 0/0/0/1/0 gvsd CLAIMED INTERFACE HPVM NPIV Stor Adapter
/dev/gvsd0
Regards,
Dave
[Any personal opinions expressed are mine, and not official statements on behalf of Hewlett Packard Enterprise]
The VM guest can see and instantiate the NPIV adapters, so that looks good. I'm a little confused by the hpvmstatus output you posted previously:
hba avio_stor 1 5 npiv /dev/fcd0-0x50060B00006499A8,0x50060B00006499A9
hba avio_stor 2 0 npiv /dev/fcd0-0x100000110A030000,0x100000110A030001
Neither of those NPIV virtual HBAs are using the GUID manager's defined range. Are you no longer using the GUID manager to manage WWNs? Did you hard-code those specific WWNs on the command-line when you added the NPIV instances to this VM?
Compare this against my VSP:
# guidmgmt -L wwn
Range type Start(Hex) End(Hex) Size(Dec)
=============== ================== ================== ==========
HP-DEFINED-PWWN 0x50014c2000000000 0x50014c2000007fff 32768
HP-DEFINED-NWWN 0x50014c2800000000 0x50014c2800007fff 32768
You can see that the *port* WWN range begins at 0x50014c2000000000 and the *node* WWN range begins at 0x50014c2800000000. Each time a virtual HBA is assigned WWNs by the GUIDmgr it assigns one WWN from each range, so my virtual servers end up with virtual HBAs looking like this:
hba avio_stor 0 0 npiv /dev/fcd0-0x50014C2000000002,0x50014C2800000002
hba avio_stor 0 1 npiv /dev/fcd1-0x50014C2000000003,0x50014C2800000003
The first number is the port WWN, and you can see it falls into the configured range on the GUIDmgr. Same goes for the node WWN value falling into the node range.
Your VM guest is not only using values that are not in the HP defined range, your VM is using *consecutive* values for the port and node WWN. Are you absolutely certain this is how you want to manage your WWNs?
Dave
[Any personal opinions expressed are mine, and not official statements on behalf of Hewlett Packard Enterprise]
The whole point of using the GUIDmgr is to have it manage the WWNs so you don't accidentally assign the same WWNs to multiple virtual servers. If you're going to hard code values then don't bother with the GUIDmgr, but you'd better be absolutely certain you know what you're doing because using the wrong values can lead to multiple servers (physical or virtual) accessing the same storage and even data corruption.
My recommendation would be to shutdown the VM, delete the existing NPIV vHBAs and create new ones using syntax similar to the following:
# hpvmmodify -P vm001 -a hba:avio_stor::npiv:/dev/fcd0
That syntax tells the VSP to assign the bus/device/function values and tells the GUIDmgr to assign the WWNs. If you want multiple vWWNs then repeat the command to create a second instance. If you have multiple physical FC ports and you want the VM to use multiple paths then you can create the second vHBA and specify a different physical FC port such as /dev/fcd1.
Dave
[Any personal opinions expressed are mine, and not official statements on behalf of Hewlett Packard Enterprise]