Hi everyone,
 
I am currently using RHOSP 16.2.4 and launched a virtual machine (VM) using a volume.
From the attach information, I confirmed that this volume was mounted on ports 1 and 2 through HBA1 and HBA2 on the compute node.
 
When I checked the devices on the compute node, I found two devices, which is as expected.
[root@overcloud-novacompute-0 heat-admin]# multipath -ll
LUN-UUID dm-5 FUJITSU,ETERNUS_DXL
size=10G features='1 queue_if_no_path' hwhandler='1 alua' wp=rw
`-+- policy='service-time 0' prio=10 status=active
  |- 3:0:2:1 sdl 8:176 active ready running
  `- 5:0:5:1 sdv 65:80 active ready running
[root@overcloud-novacompute-0 heat-admin]# ll /dev/disk/by-path/
lrwxrwxrwx. 1 root root  9 Aug 20 08:51 fc-HBA1-port1-lun-0 -> ../../sdv
lrwxrwxrwx. 1 root root  9 Aug 20 08:51 pci-0000:00:0a.0-fc-port1-lun-0 -> ../../sdv
lrwxrwxrwx. 1 root root  9 Aug 20 08:51 fc-HBA2-port2-lun-0 -> ../../sdl
lrwxrwxrwx. 1 root root  9 Aug 20 08:51 pci-0000:00:0b.0-fc-port2-lun-0 -> ../../sdl

I then shut down the VM and the compute node, and replaced HBA1 with HBA3 on the compute node.
Under these circumstances, the volume should be mounted on ports 1 and 2 through HBA3 and HBA2 respectively.
In other words, I think I should be able to find the following devices.
device1: fc-HBA3-port1-lun-0 -> ../../sdl
                   pci-0000:00:0c.0-fc-port1-lun-0 -> ../../sdl
device2: fc-HBA2-port2-lun-0 -> ../../sdv
                   pci-0000:00:0b.0-fc-port2-lun-0 -> ../../sdv

However, after restarting the compute node and the VM,
although the connection and read/write operations are functioning normally,
the compute node fails to find the volume on HBA3.
[root@overcloud-novacompute-0 heat-admin]# multipath -ll
LUN-UUID dm-5 FUJITSU,ETERNUS_DXL
size=10G features='1 queue_if_no_path' hwhandler='1 alua' wp=rw
`-+- policy='service-time 0' prio=10 status=enabled
  `- 5:0:5:1 sdl 8:176 active ready running
[root@overcloud-novacompute-0 heat-admin]# ll /dev/disk/by-path/
lrwxrwxrwx. 1 root root  9 Aug 20 08:51 fc-HBA2-port2-lun-0 -> ../../sdl
lrwxrwxrwx. 1 root root  9 Aug 20 08:51 pci-0000:00:0b.0-fc-port2-lun-0 -> ../../sdl

Could you please advise on how I can make the compute node find the volume on HBA3 in this scenario?
And I would like to know how to change the Initiator from HBA1 to HBA3 in the initiator_target_map of DB without detach volume?
 
Regards,
Inori.