Installing HP NC523SFP 10Gbps Ethernet on CentOS 8

[nroskam]

Hi,

I'm upgrading my HP ML10 v2 Proliant server from a 4x1Gb NIC, to a 2x10Gb NIC. I use the server as an all-purpose home server, e.g. Unifi Controller, NAS, Plex, File Server (Samba), TimeMachine (Netatalk), VPN, some KVM VMs, etc, etc.

The current HP NC365T aggregated 4-port 1Gbps is well supported, and I got it to work easily when I switched to CentOS 8 (from 7) recently in a bonded setup.

The new HP NC523SFP 2-port 10Gbps SFP+ seems a bit harder to get to work on my CentOS 8 install? Did I even read that the standard driver for it may have been deprecated?

I run CentOS 8.1.1 and kernel 4.18.0-147.5.1.el8_1.x86_64.

Code: Select all

Linux hpnas 4.18.0-147.5.1.el8_1.x86_64 #1 SMP Wed Feb 5 02:00:39 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux

lspci | grep Ethernet recognises the onboard NIC (2x Broadcom), the NC365T (4x Intel) and the NC523SFP (2x Qlogic):

Code: Select all

03:00.0 Ethernet controller: Broadcom Inc. and subsidiaries NetXtreme BCM5720 2-port Gigabit Ethernet PCIe
03:00.1 Ethernet controller: Broadcom Inc. and subsidiaries NetXtreme BCM5720 2-port Gigabit Ethernet PCIe
04:00.0 Ethernet controller: QLogic Corp. cLOM8214 1/10GbE Controller (rev 54)
04:00.1 Ethernet controller: QLogic Corp. cLOM8214 1/10GbE Controller (rev 54)
0a:00.0 Ethernet controller: Intel Corporation 82580 Gigabit Network Connection (rev 01)
0a:00.1 Ethernet controller: Intel Corporation 82580 Gigabit Network Connection (rev 01)
0a:00.2 Ethernet controller: Intel Corporation 82580 Gigabit Network Connection (rev 01)
0a:00.3 Ethernet controller: Intel Corporation 82580 Gigabit Network Connection (rev 01)

The Qlogic NICs are not recognised when running nmcli:

Code: Select all

br0: connected to br0
        "br0"
        bridge, xx:xx:xx:xx:xx:xx, sw, mtu 1500
        ip4 default
        inet4 192.nn.nn.nn/24
        route4 192.nn.nn.nn/24
        route4 0.0.0.0/0
        inet6 nnnn::nnnn:nnnn:nnnn:nnnn/64
        route6 nnnn::/64
        route6 nnnn::/8

bond0: connected to bond0
        "bond0"
        bond, xx:xx:xx:xx:xx:xx, sw, mtu 1500
        master br0

ens2f0: connected to ens2f0
        "Intel 82580"
        ethernet (igb), xx:xx:xx:xx:xx:xx, hw, mtu 1500
        master bond0

ens2f1: connected to ens2f1
        "Intel 82580"
        ethernet (igb), xx:xx:xx:xx:xx:xx, hw, mtu 1500
        master bond0

ens2f2: connected to ens2f2
        "Intel 82580"
        ethernet (igb), xx:xx:xx:xx:xx:xx, hw, mtu 1500
        master bond0

ens2f3: connected to ens2f3
        "Intel 82580"
        ethernet (igb), xx:xx:xx:xx:xx:xx, hw, mtu 1500
        master bond0

eno1: disconnected
        "Broadcom NetXtreme BCM5720"
        ethernet (tg3), xx:xx:xx:xx:xx:xx, hw, mtu 1500

eno2: disconnected
        "Broadcom NetXtreme BCM5720"
        ethernet (tg3), xx:xx:xx:xx:xx:xx, hw, mtu 1500

lo: unmanaged
        "lo"
        loopback (unknown), 00:00:00:00:00:00, sw, mtu 65536

DNS configuration:
        servers: 192.nn.nn.nn
        interface: br0

Does anyone have any suggestions on how to get this to work? I can't seem to find much for it online?

Thx

[TrevorH]

You forgot to add the -nn switch to lspci so I am guessing that the PCI vendor/device id for this is 1077:8020. Running grep 1077 /lib/modules/4.18.0-147.5.1.el8_1.x86_64/modules.* | grep 8020 returns nothing so it does look like this device is not supported by the CentOS kernel.

I checked in ELRepo and I do not see a kmod-qlogic or another kmod that I instantly recognise as being for this device but you could raise an RFE in the ELRepo bug tracker and ask them if it's something they could provide.

dnf provides '*1077*8020*' --enablerepo=elrepo-testing doesn't find anything either but that's not entirely reliable.