Booting Clear Linux installer from an existing PXE Server

Im have an existing network boot environment with Windows Deployment Services, that Is using So my boot config works perfectly with CentOS, Debian and Windows…

After I have seen the benchmark performance tests on Clear Linux, I wanted to give it a go.
So I added it into my existing PXE boot config.

I added the kernel.xz and initrd.gz to the PXE Server Root, and I can start the kernel.

Then the problem.

I cant find the documentation of all the variables I can use in the APPEND part of the pxeLinux.cfg .
What I really hope is to find a variable, that allows me to add a custom repository location, the same like I do with CentOS and Debian.


KERNEL vmlinuz
APPEND initrd=initrd.img ramdisk_size=100000 ip=dhcp repo= ks=
LINUX kernel.xz
INITRD initrd.gz
APPEND quiet init=/usr/lib/systemd/systemd-bootchart initcall_debug tsc=reliable no_timer_check noreplace-smp rw initrd=initrd isterconf=http://install.*****.no/icis/ister.conf

Which image did you extract kernel and initrd from?

There was as transition from an old installer (ister) to the new installer (clr-installer). The current documentation on PXE still uses the old installer (ister).

Your APPEND line looks like you’re making reference to the old ister installer but your screenshot looks like it is using the new clr-installer image.

There is an open GH issue for this where someone has shared tips on getting the clr-installer to work over PXE though.

I extracted the image from clear-30970-live-server.iso

I have an idea of using initrd= to point to the rootfs.img so i guess ill test some more =)

ok, so after multiple tests Im not able to boot the live-server installer,
and batch install ClearLinux on my render farm.

Im able to boot the ClearLinux PXE image, but that doesn’t connect to the network cars, nor will it work with multiple custom packages.

Where am I now ?

Im now in the prosess of creating myself an custom image based on a clean install using clr-installer from the .iso of Live-server. from there Ill do the install, and generate the clr-installer.yaml configuration file.

Then I will use that configuration file, to generate an installer image, that I hope will include the packages I need.


Oof sorry for the pain. A custom repo isn’t really something we support in Clear Linux at this time so I’m a little confused as to what the new installer image has you are using to handle this case. Would you mind describing the mechanism a little more?

Sure, as the PXE image did boot fine, but where missing networking drivers, and some packages we use for 3D rendering I thought that it would be possible to build that from the live-server iso that did have all this…

But unfortunately I failed at it, and as I don’t have more time on this project I went back to CentOS.

I will try more with the Azure image, as I know we will spin up around 700 machines there soon for rendering.

The reason for pushing an PXE installer similar to CentOS ‘kickstart or better is to be able to run the same OS on both cloud and local computes.

And yeah, to spice things up a bit, the PXE server is an Windows Deployment server with a pxelinux mini hack on top :wink:

On my initial tests, ClearLinux did perform 10-15% better at 3D rendering.