On VIOS vscsi configuration it should be used the existing SVC host mapping during PowerVC VM deployment.

In one configuration, PowerVC order toward SVC to create new host definition even if it is the vscsi configuration.
I think the mkvhost command should not be executed in a vscsi configuration during PowerVC VM deployment.

Detailed content document attached, "PowerVC_vscsi_deployment_problem_with_2SVCs_ideas.pptx".

# Background

The VM OS image volume consists of two LUNs from SVC1 and SVC2 with OS LVM mirror.
Storage connection group has the settings for 4 ports settings  from fcs0,2 of VIOS #1 and  VIOS #2  on PowerVC.
The connection between VIOS and VIOC is a virtual SCSI configuration. 
SVC1 defines the two VIOS hosts for fcs0 and SVC2 defines the two VIOS hosts for fcs2.
 

# issue

In this case,  it results in the VM deployment failure and it behaves like creating a new host for fcs0 on SVC2, and creating a new host for fcs2 on SVC1.
(see attached pptx file page 1)

To avoid this phenomenon, VIOS fcs2, which does not configured on fabrics was pre-added to the VIOS host on SVC1 and VIOS fcs0, which does not configured on fabrics, was pre-added to the VIOS host on SVC2 and the VM deployment was succeeded in that configuration. However, the VIOS host shows degraded status in SVCs.   
(see attached pptx file page 2,3 )

# point for improvement

The mkvhost command should not be executed in a vscsi configuration and be used the existing SVC host mapping during PowerVC VM deployment.