Adding / Troubleshooting FCoE storage (HPe Synergy)

[PoudNL]

I am quite new to Talos Linux but really loving it already.

Currently I am designing a new bare-metal Kubernetes cluster that eventually will also be our next virtualization platform with Kubevirt. Using Talos Linux as the OS would be great, but I am limited in hardware choices and need to use HPe Synergy Compute nodes with Fibre Channel over Ethernet (Qlogic). No local disks are installed, bootdisk comes from a SAN (boot from SAN). Which works fine with a regular Ubuntu installation or VMware installation.

I tried installing Talos Bare-metal 1.7.6 (and 1.8.0), but had the issue that my networking and storage were not detected. After adding the qlogic-extension, networking started to work seamless but the SAN disk is still not available.
The following command just did not give any output:

talosctl --nodes <node-ip> --endpoints <endpoint-ip> disks -i

When I apply a config to the node, the dashboard shows the error that disk /dev/sda is not found and installation fails.

Hardware info

• Synergy 480 Gen10 v2
• Synergy 4820C 10/20/25Gb CNA / QLogic FastLinQ QL45xxx Series 10/25 GbE Controller (FCoE)

qedf drivers
I investigated the drivers that Ubuntu uses and these are the qedf modules. I think these are part of the linux-firmware qed folder, which is used in the qlogic extension.
So I was hoping that this extension would be enough to make it work, but it didn't.

Building custom extension
I also tried looking into the procedure to create a custom extension and building a custom extension where also the qlogic was added to the qlogic-firmware extension, following this guide, but the documentation is just to 'rough' for a newbie like me. After building I just don't know how to continue.

Building custom ISO
Also tried following Imager to create my own ISO with the new builded extension, but I just don't know how to bring the two together (add my custom extension to the ISO).

So now I am a bit clueless how to continue my investigation / troubleshooting.
Maybe it is some Fibre channel module that is missing or some other driver, I just don't know how to start.

Any help is very much appreciate :)

[smira]

Any information from Ubuntu or any other distro might be helpful.

From what you're saying it should be https://cateee.net/lkddb/web-lkddb/QEDF.html ? We can get it enabled for Talos 1.8.1

[smira]

Thanks, this is very detailed, and enough for me to get started. I will do a PR and backport it to 1.8.x.

[PoudNL]

I hope you (or someone else) can also share a bit how others can work on this.
Documentation is not really (for me at least) clear enough to help adding drivers/modules in the future.

[smira]

Looks like every driver/module is available either as builtin or as a module, except for qedf, I'm going to enable that.

[smira]

siderolabs/pkgs#1046 is a part number 1

[smira]

#9436 is part number 2 (and final one)

This will be backported to Talos 1.8.1

[PoudNL]

I just tested 1.8.1 with the qlogic extension on our hardware and I can confirm that this fixed my issue! I can see storage devices that are presented over FCoE and was able to install Talos on these disks.

Thanks again for the fast implementation!