osbuild-composer
osbuild-composer copied to clipboard
osbuild
Support ignition in simplified-installer and raw-image
This patchset is the basic work needed to support ignition in r4e - it doesn't include any blueprint or file injection change as I wanted to maintain it easy to review and since this won't break anything, I think it'll be beneficial to merge this before other work to plumb ignition files/config in the blueprints. This PR also includes a commit that enables kargs injection from the installer and raw-image blueprint to effectively enable passing the artifact an ignition.config.url and test it out. Lastly, to make the work bound to the sysroot.ro=true patch, this also includes https://github.com/osbuild/osbuild-composer/pull/3053
Things missing:
- the
ignition-edgeRPM isn't yet packaged anywhere (working with the coreos team to ship it), the source lives https://github.com/fedora-iot/ignition-edge - https://gitlab.com/redhat/centos-stream/rpms/ignition/-/merge_requests/13 - polish commits
- the work here is rhel 9 only ~~, I'll port it to 8 too once we have
ignition-edge(???)~~ - ~~add a basic integration test for raw-image and simplified-installer~~
Picking this up to rebase it after #3166
(@runcom there is a copy of your original code on https://github.com/7flying/osbuild-composer/tree/ignition-poc-from-runcom, just in case)
- drop https://github.com/7flying/osbuild-composer/commit/59291dbb8ff9c6482362aabd3ef9bd9a47c7c24b: already done by #3178
~~Messed up somewhere during the rebase and got a detached head, see https://github.com/osbuild/osbuild-composer/pull/3189~~
- Rebase required changes on https://github.com/osbuild/osbuild-composer/pull/3130/commits/dfa53af87747d9285e1941cfd72dd724f1102352
yeah, this is the actual ignition enablement but we first need a new 9.2 snapshot and also the 9.2 runners (but don't take my word, I've catching up after PTO)
D'oh! I should not have merged PR #3161 first, that's sort of misleading as it is now. @ondrejbudai should I revert the other merged PR? I guess these two should have been combined into one PR.
Is this PR still relevant?
Yes, this is the counterpart for #3161 to do something with the ignition stuff passed in the blueprints into the ISO.
D'oh! I should not have merged PR #3161 first, that's sort of misleading as it is now. @ondrejbudai should I revert the other merged PR? I guess these two should have been combined into one PR.
It is fine, #3161 does not break anything, it puts things into the ISO. We do not do anything with those since this PR is missing, and that's it.
We divided the work hence two PRs, it did make sense to us :sweat_smile:
I've looked through this while it was being prepared but didn't look too closely since it kept changing and was marked as draft. I see it's green now, and has been for a couple of weeks. Is it ready to review?
I've looked through this while it was being prepared but didn't look too closely since it kept changing and was marked as draft. I see it's green now, and has been for a couple of weeks. Is it ready to review?
it is ready to review but it's still dependent on https://github.com/osbuild/osbuild-composer/pull/3134
I'd also like a more descriptive commit message on the last commit ("wire ignition").
that's one last WIP I have - I'll polish everything up for a final review anyway
tests are green - I'm gonna polish this all tomorrow, pending a tiny decision about ignition.embedded.url and adding a test for raw image (just ignition.firstboot)
Also, in the kernel options in the manifests I'm seeing ...modprobe.blacklist=vc4,rw,,coreos.no_persist_ip,... (an empty option between rw and coreos.no_persist_ip) but I can't pinpoint where it's coming from.
@achilleas-k I found it, comes from customization.GetKernel() and how we append - basically checking for nil there isn't sufficient cause it's never nil - I need to check that Append isn't empty before appending