Proxmox Lab Build | Michael Krawczyk

Bare-Metal Install of Proxmox VE on HPE DL380 G7

Download the latest Proxmox VE ISO from the official site. Write it to USB (Rufus, Ventoy, Etcher, or dd). Boot the DL380 from USB, select "Install Proxmox VE", choose local disks (keep this separate from the MSA), set root password and management IP on your LAN — never on the iSCSI subnet. After install, log in at https://<mgmt-ip>:8006.

Foundation

Update Proxmox and Adjust Repositories

On a lab without a subscription, swap the enterprise repo for the no-subscription repo, then fully patch the host before touching storage or drivers.

// switch repos + upgrade + reboot

sed -i 's/enterprise/no-subscription/' /etc/apt/sources.list.d/pve-enterprise.list
echo "deb http://download.proxmox.com/debian/pve bookworm pve-no-subscription" \
  > /etc/apt/sources.list.d/pve-no-sub.list

apt update && apt full-upgrade -y
reboot

Patch First

Install Dependencies and Verify NICs

On older hardware like the G7, confirm NIC visibility and install iSCSI tooling before layering any storage config on top. Install firmware packages if NIC or HBA behavior is abnormal.

// install iscsi + multipath, confirm interfaces

apt install -y open-iscsi multipath-tools firmware-linux-free firmware-linux-nonfree
systemctl enable --now open-iscsi

ip a
lspci | grep -i ethernet
# target NICs: ens2f0 (→ Ctrl B .51) and ens3f1 (→ Ctrl A .50)

Verify First

Design the iSCSI Network Layout Before Touching Interfaces

Map out IPs, cabling, and routing rules before editing a single config file. The critical rules: iSCSI NICs carry no default gateway, /32 host routes keep traffic on the correct interface, and rp_filter is relaxed to prevent Linux from dropping asymmetric return traffic.

Interface	IP	Route To	Purpose
ens2f0	192.168.1.40/24	→ MSA Ctrl B (.51)	iSCSI Path 1
ens3f1	192.168.1.41/24	→ MSA Ctrl A (.50)	iSCSI Path 2
MSA Ctrl A	192.168.1.50	—	Primary Controller
MSA Ctrl B	192.168.1.51	—	Secondary Controller

Plan It Out

Configure iSCSI NICs in /etc/network/interfaces

With the layout locked, apply static IPs, per-interface host routes, and rp_filter settings. After writing the file, restart networking or reboot, then validate every route lands on the expected NIC.

// /etc/network/interfaces — iSCSI NIC config

auto ens2f0
iface ens2f0 inet static
    address 192.168.1.40/24
    post-up ip route replace 192.168.1.51/32 dev ens2f0 src 192.168.1.40
    post-up sysctl -w net.ipv4.conf.ens2f0.rp_filter=2

auto ens3f1
iface ens3f1 inet static
    address 192.168.1.41/24
    post-up ip route replace 192.168.1.50/32 dev ens3f1 src 192.168.1.41
    post-up sysctl -w net.ipv4.conf.ens3f1.rp_filter=2

// validate — both pings and routes must pass

ping 192.168.1.50  # → should reply
ping 192.168.1.51  # → should reply
ip route get 192.168.1.50  # → must show ens3f1
ip route get 192.168.1.51  # → must show ens2f0

Wire It Up

Configure the HPE MSA 2050 — LUN, Pool, and Host Mapping

On the MSA side: confirm controller IPs, subnet mask, and link health. Create your RAID-backed disk group and a ~2TB volume. Then create a host entry using the Proxmox iSCSI initiator IQN and map the LUN to that host. The IQN must match exactly what Proxmox reports — one character off and discovery returns nothing.

// get proxmox initiator IQN to use on MSA

cat /etc/iscsi/initiatorname.iscsi
# copy this value into MSA host entry verbatim

SAN Side

iSCSI Discovery and Login — Bring the LUN Online

Discover targets from both controllers, then log in. If discovery hangs, the problem is almost always routing or rp_filter — go back to Step 5. Once logged in, the kernel should report a new block device. Confirm with dmesg and lsblk before touching LVM.

// discover + login + verify block device

iscsiadm -m discovery -t sendtargets -p 192.168.1.50
iscsiadm -m discovery -t sendtargets -p 192.168.1.51
# both should return the MSA IQN

iscsiadm -m node --login

dmesg | tail       # look for new scsi device
lsblk              # new /dev/sdX should appear
fdisk -l           # confirm 2TB size

Go Live

Build LVM Volume Group and Register devlab in Proxmox

Initialize the iSCSI block device as a physical volume, create the vg_iscsi volume group, then wire it into the Proxmox GUI as the devlab storage target. Future expansion is non-destructive: extend the LUN on the MSA and resize LVM in place.

// LVM setup — replace sdX with your device

pvcreate /dev/sdX
vgcreate vg_iscsi /dev/sdX
vgs && pvs
# should show ~2TB free in vg_iscsi

In the Proxmox web UI: Datacenter → Storage → Add → LVM. Set ID to devlab, volume group to vg_iscsi, enable for disk images and containers. Save — devlab is now live and backed by the MSA 2050.

Done