Boot Luns Fiber Channel (FC)

[douglasvital]

I have 3 hosts that I installed Proxmox, the boot is with Fiber Channel (FC)

After installing proxmox, I installed multipath-tools and put another three Luns of 1 TB each via FC.

At the command output

`````
multipath -ll
````

The normal DS appears, but the boot LUN does not appear as an active multipah, it is only on one side of the controller.
Being able to place the normally shared DS on all hosts via FC

The problem is that at boot time the multipath is not regulated and it is only on one side of the Storage someone. Knows how to agree.






root@proxmoxm3-01:~# lsblk -f
NAME FSTYPE FSVER LABEL UUID FSAVAIL FSUSE% MOUNTPOINTS
sda
├─sda1
├─sda2 vfat FAT32 DD34-7F50
└─sda3 LVM2_member LVM2 001 EH8R3C-nPll-2Xb3-6hIn-pSWB-DPcv-IUwUq1
sdb
├─sdb1
├─sdb2 vfat FAT32 DD34-7F50
└─sdb3 LVM2_member LVM2 001 EH8R3C-nPll-2Xb3-6hIn-pSWB-DPcv-IUwUq1
sdc LVM2_member LVM2 001 HhZ9GA-ub5O-qLAE-mi9h-8v6e-G35C-lhGtXi
└─mpathb LVM2_member LVM2 001 HhZ9GA-ub5O-qLAE-mi9h-8v6e-G35C-lhGtXi
└─vg_ds_8200-vm--100--disk--0
sdd LVM2_member LVM2 001 HhZ9GA-ub5O-qLAE-mi9h-8v6e-G35C-lhGtXi
└─mpathb LVM2_member LVM2 001 HhZ9GA-ub5O-qLAE-mi9h-8v6e-G35C-lhGtXi
└─vg_ds_8200-vm--100--disk--0
sde LVM2_member LVM2 001 id71E0-nHp0-r72H-bjjc-ZIF9-GDUx-THsFki
└─mpathc LVM2_member LVM2 001 id71E0-nHp0-r72H-bjjc-ZIF9-GDUx-THsFki
sdf LVM2_member LVM2 001 id71E0-nHp0-r72H-bjjc-ZIF9-GDUx-THsFki
└─mpathc LVM2_member LVM2 001 id71E0-nHp0-r72H-bjjc-ZIF9-GDUx-THsFki
sdg mpath_member
└─mpathd
sdh mpath_member
└─mpathd
sdi
├─sdi1
├─sdi2 vfat FAT32 DD34-7F50
└─sdi3 LVM2_member LVM2 001 EH8R3C-nPll-2Xb3-6hIn-pSWB-DPcv-IUwUq1
├─pve-swap swap 1 25703b09-5792-4d11-ab3b-64297a12942b [SWAP]
├─pve-root xfs 82ca20d3-3ba0-4fd8-b84d-79b8f27da774 14.3G 52% /
├─pve-data_tmeta
│ └─pve-data
└─pve-data_tdata
└─pve-data
sdj
├─sdj1
├─sdj2 vfat FAT32 DD34-7F50
└─sdj3 LVM2_member LVM2 001 EH8R3C-nPll-2Xb3-6hIn-pSWB-DPcv-IUwUq1
sdk LVM2_member LVM2 001 HhZ9GA-ub5O-qLAE-mi9h-8v6e-G35C-lhGtXi
└─mpathb LVM2_member LVM2 001 HhZ9GA-ub5O-qLAE-mi9h-8v6e-G35C-lhGtXi
└─vg_ds_8200-vm--100--disk--0
sdl LVM2_member LVM2 001 HhZ9GA-ub5O-qLAE-mi9h-8v6e-G35C-lhGtXi
└─mpathb LVM2_member LVM2 001 HhZ9GA-ub5O-qLAE-mi9h-8v6e-G35C-lhGtXi
└─vg_ds_8200-vm--100--disk--0
sdm LVM2_member LVM2 001 id71E0-nHp0-r72H-bjjc-ZIF9-GDUx-THsFki
└─mpathc LVM2_member LVM2 001 id71E0-nHp0-r72H-bjjc-ZIF9-GDUx-THsFki
sdn LVM2_member LVM2 001 id71E0-nHp0-r72H-bjjc-ZIF9-GDUx-THsFki
└─mpathc LVM2_member LVM2 001 id71E0-nHp0-r72H-bjjc-ZIF9-GDUx-THsFki
sdo mpath_member
└─mpathd
sdp mpath_member
└─mpathd
root@proxmoxm3-01:~#




root@proxmoxm3-01:~# cat /etc/fstab
# <file system> <mount point> <type> <options> <dump> <pass>
/dev/pve/root / xfs defaults 0 1
/dev/pve/swap none swap sw 0 0
proc /proc proc defaults 0 0

 

[bbgeek17]

Proxmox ISO/installation does not support boot multipath, if I remember correctly. You need to use original Debian ISO and ensure that multipath is specified during the installation.

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[douglasvital]

Proxmox ISO/installation does not support boot multipath, if I remember correctly. You need to use original Debian ISO and ensure that multipath is specified during the installation.

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

I understand, what a shame, it was a good option in the company I work at, we only have FC as a boot.

Thanks for the answer, I'll try Ovirt

 

[bbgeek17]

I understand, what a shame, it was a good option in the company I work at, we only have FC as a boot.

Installing Debian and then Proxmox packages is a supported solution by PVE: https://pve.proxmox.com/wiki/Install_Proxmox_VE_on_Debian_12_Bookworm

Thanks for the answer, I'll try Ovirt

Sure, just make sure you have a migration plan for 2026 EOL

Good luck

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[douglasvital]

Installing Debian and then Proxmox packages is a supported solution by PVE: https://pve.proxmox.com/wiki/Install_Proxmox_VE_on_Debian_12_Bookworm


Sure, just make sure you have a migration plan for 2026 EOL

Good luck

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

Ovirt has no EOL plans for 2026, Redhat will no longer be the master contributor, but Ovirt will continue, even Oracle is joining the project with OLVM which has indefinite support.

about Proxmox I tried to contact support to understand if FC boot is supported natively and they didn't answer me, only if I buy the license, but I only wanted one yes is supported because then I would proceed with the installation and test the product and then with I would definitely buy the enterprise support.

 

[bbgeek17]

Ovirt has no EOL plans for 2026, Redhat will no longer be the master contributor, but Ovirt will continue, even Oracle is joining the project with OLVM which has indefinite support.

I should have been more clear: Redhat will stop enterprise support for Red Hat Virtualization product (Ovirt based). Of course, as an open source product, Ovirt will continue. Thanks for pointing it out.

https://access.redhat.com/announcements/6960518

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[douglasvital]

I should have been more clear: Redhat will stop enterprise support for Red Hat Virtualization product (Ovirt based). Of course, as an open source product, Ovirt will continue. Thanks for pointing it out.

https://access.redhat.com/announcements/6960518

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

Thank you for your help.

 

[LnxBil]

The normal DS appears, but the boot LUN does not appear as an active multipah, it is only on one side of the controller.
Being able to place the normally shared DS on all hosts via FC

Is this shown in the output and I'm blind? If not, please try to see in the debug multipath output why it is blacklisted.

I haven't done this with PVE yet, but I'd suspect it would work as it would on older RHEL. You can always build you own multipath-enabled initrd so that it gets support on boot. I have no idea how to do this, yet it would probably work. I don't have a working FC-based env at my handy right now, yet I can try to simulate it within PVE itself.

 

[LnxBil]

... yet I can try to simulate it within PVE itself.

I wasn't able to simulate it properly, even the Debian install did not recognize it correctly: multipath -v4 yielded

sda: blacklisted, udev property missing