On Thu, May 6, 2021 at 10:47 PM Yasha Karant <[email protected]> wrote: > > From below: > > I get a consistent build with any of the three OS every time. > > Is Kickstart a mechanism to build (from source) an EL8 image that is > then bootable and installable?
You *can* use it to build a base installation, or an image that can be read and replicated in virtualization deployment systems. I've done so with roughly..... 20,000 systems in my career, thoush most of them installed an OS from a much smaller tarball on arbitrarily partitioned systems. It's basically what docker does with tarballs, bundling an OS image. > My understanding is that using Kickstart, one could arrange for the > appropriate disk partitions (including those such as /home or /opt that > should not be touched by any new install, merely made accessible under > the new OS), and the installation of the new release RPMs into the new > target OS (e.g., generate a kickstart in SL7 and then modify that > kickstart 7 using the RL described application to kickstart 8 that would > correctly install, say, AlmaLinux 8, from the "binary executable" RPMs, > not SRPMs) without the issues that have been described for the current > EL8 production Anaconda. Well, yes. Look up "PXE" and "kickstart" for more details on how to tune these when you have control of the DHCP setup. In particular, DHCP can support a PXE server that lists a variety of available, selectable options for network boots, including just proceeding with the hard-disk boot, providing a recovery network boot, or base-installing an OS. Those can also be tuned from the DHCP and PXE server to provide pe-configured defaults. > Am I missing something? Only that it takes time and work to set up. It's what competent system engineers did before paying for AWS or VMWare services to do much of the work, especially setting up gold images, for you.
