dracut: adapt for s/oem/platform in Ignition #45
Conversation
cf21917 to
3979dff
Compare
|
We're not building ignition rpms from |
Wouldn't it be cleaner to just match the Ignition repo and use a |
|
Yeah that's probably better. Since the RPM spec file refers to it by commit that should be fine. |
|
But we should create that branch and add notes to the readmes before merging this |
|
OK, in favour of splitting. Though for this patch, I'd rather we maintain compatiblity for |
3979dff to
3068bb3
Compare
With coreos/ignition#735 merged, we need to fix our service units to use `--platform` here. Also just do the equivalent thing here and start to move away from the oem terminology. Crucially though, we still look for `coreos.oem.id` until downstreams have adapted (notably coreos-assembler). Closes: coreos#31
3068bb3 to
67e092e
Compare
|
this LGTM - re: spec2x I'd honestly rather just keep a single branch here and move to 3.0 spec sooner than later. For RHCOS maybe we could just backport patches in the rpm? |
only for ones that are absolutely required, of course |
|
OK, this is tested now, both using the old and the new kernel args. Dropping WIP, though let's see what people think re. |
+1 for |
discussed in IRC. We'll go with this for now: |
|
FYI: created the spec2x branch: https://github.com/coreos/ignition-dracut/tree/spec2x |
There was a problem hiding this comment.
ignition.platform.id makes sense in the context of this repo, which is intended to be distro-independent. But for FCOS, at least, we'll be configuring a number of other distro behaviors (whether to start a platform agent, etc.) on the "Ignition" platform ID. That might make the name confusing. I don't have a better suggestion, though.
| @@ -95,4 +98,4 @@ if [[ $(systemd-detect-virt || true) =~ ^(kvm|qemu)$ ]]; then | |||
| oem_id=qemu | |||
| fi | |||
|
|
|||
| echo "OEM_ID=$(cmdline_arg coreos.oem.id ${oem_id})" > /run/ignition.env | |||
| echo "PLATFORM_ID=$platform_id" > /run/ignition.env | |||
There was a problem hiding this comment.
This breaks the oem_id logic in the previous stanza.
There was a problem hiding this comment.
Good catch! Added a fix for this in #46.
Minor regression from coreos#45.
With coreos/ignition#735 merged, we need to fix
our service units to use
--platformhere. Also just do the equivalentthing here and start to move away from the oem terminology. Crucially
though, we still look for
coreos.oem.iduntil downstreams haveadapted (notably coreos-assembler).