CentOS 8 support is now available as a third Qlustar edge-platform and
was published together with the latest security updates. It comes with
the same functionality as our CentOS 7 port which includes out of the
box OpenHPC 1.3.9 integration, Slurm 18.10.x, up-to-date Nvidia 340.x
drivers, Lustre client and BeeGFS client.
Making your compute nodes switch to CentOS 8 on an existing Qlustar
cluster installation is extremely simple: First create a CentOS 8 chroot
 optionally adding OpenHPC base support, then generate a CentOS 8 OS
image  selecting the modules you require and finally a boot config
 using the just generated image. After assigning  this boot config
to the nodes you want to run it on, you just need to reboot them and
voilà your CentOS 8 nodes are ready to be used.
Note that OpenHPC 1.3.9 was originally built for CentOS 7.7. We had to
make a few adjustments to make the OpenHPC parts relevant to Qlustar
(compilers, MPIs, library packages and applications) work for
CentOS 8. The OpenHPC test-suite was run successfully against these
components. Once OpenHPC 2.0 will be out, we'll publish instructions to
update to it. This will be straightforward.
Additional improvements with this release: Updates to the latest Lustre
2.12.4 LTS release and the newest Nvidia graphics drivers 340.64 adding
support for the latest Nvidia GPUs. These two updates apply to all
Qlustar 11.0 platforms (Ubuntu 18.04 and CentOS 7/8).
There are six relevant network interfaces on my head node, some of which are relevant and linked and some of which are not, and I can't tell which is which from the network interface names and am not on site to plug and unplug things. Is there some way to shell out, bring all the interfaces up and puzzle it out that way? TIA.