My customer requested to configure a VM in Proxmox in the way similar it is in POWER (VIOS/AIX) systems, mean, a virtual FC host bus adapter should be created over two physical links with it's own wwn that should be transparently presented to FC switching fabric. I've found only few pages where something similar is mentioned. One of them is access.redhat.com/documentation/en-us/red_hat_enterprise_linux/7/html/virtualization_deployment_and_administration_guide/sect-npiv_storage.
It is expected that vHBA assigned to VM will move together with VM to another physical server in cluster in case of failure and following halt of the running server.
Volume from storage should go via vHBA stright to vm and it's presence should not be visible on host in any form.
Root of such desire is to avoid mess with multipath configuration. Basically host server will keep more than 10 VM's where each VM will have at least one volume from storage presented over FC. Multipathing then will be configured on every VM, and link voting in case of something would be quick. If all volumes are presented to host and then forwarded to VM, path change becomes too long and VM's start to report lost volumes.
Does proxmox support such functionality? Or anything similar?
It is expected that vHBA assigned to VM will move together with VM to another physical server in cluster in case of failure and following halt of the running server.
Volume from storage should go via vHBA stright to vm and it's presence should not be visible on host in any form.
Root of such desire is to avoid mess with multipath configuration. Basically host server will keep more than 10 VM's where each VM will have at least one volume from storage presented over FC. Multipathing then will be configured on every VM, and link voting in case of something would be quick. If all volumes are presented to host and then forwarded to VM, path change becomes too long and VM's start to report lost volumes.
Does proxmox support such functionality? Or anything similar?
