Examples showing `#cloud-config` files with `stages.boot` sections don't work

[kruton]

Kairos version:

PRETTY_NAME="Ubuntu 24.04 LTS"
NAME="Ubuntu"
VERSION_ID="24.04"
VERSION="24.04 LTS (Noble Numbat)"
VERSION_CODENAME=noble
ID=ubuntu
ID_LIKE=debian
HOME_URL="https://www.ubuntu.com/"
SUPPORT_URL="https://help.ubuntu.com/"
BUG_REPORT_URL="https://bugs.launchpad.net/ubuntu/"
PRIVACY_POLICY_URL="https://www.ubuntu.com/legal/terms-and-policies/privacy-policy"
UBUNTU_CODENAME=noble
LOGO=ubuntu-logo
KAIROS_VERSION="v3.1.3-v1.31.0-k3s1"
KAIROS_PRETTY_NAME="kairos-standard-ubuntu-24.04 v3.1.3-v1.31.0-k3s1"
KAIROS_FLAVOR="ubuntu"
KAIROS_RELEASE="v3.1.3"
KAIROS_REGISTRY_AND_ORG="quay.io/kairos"
KAIROS_BUG_REPORT_URL="https://github.com/kairos-io/kairos/issues"
KAIROS_GITHUB_REPO="kairos-io/kairos"
KAIROS_NAME="kairos-standard-ubuntu-24.04"
KAIROS_VERSION_ID="v3.1.3-v1.31.0-k3s1"
KAIROS_IMAGE_LABEL="24.04-standard-amd64-generic-v3.1.3-k3sv1.31.0-k3s1"
KAIROS_ARTIFACT="kairos-ubuntu-24.04-standard-amd64-generic-v3.1.3-k3sv1.31.0+k3s1"
KAIROS_FLAVOR_RELEASE="24.04"
KAIROS_SOFTWARE_VERSION_PREFIX="k3s"
KAIROS_ID="kairos"
KAIROS_IMAGE_REPO="quay.io/kairos/ubuntu:24.04-standard-amd64-generic-v3.1.3-k3sv1.31.0-k3s1"
KAIROS_VARIANT="standard"
KAIROS_MODEL="generic"
KAIROS_TARGETARCH="amd64"
KAIROS_HOME_URL="https://github.com/kairos-io/kairos"
KAIROS_SOFTWARE_VERSION="v1.31.0+k3s1"
KAIROS_ID_LIKE="kairos-standard-ubuntu-24.04"
KAIROS_FAMILY="ubuntu"

CPU architecture, OS, and Version:

Linux beta 6.8.0-44-generic #44-Ubuntu SMP PREEMPT_DYNAMIC Tue Aug 13 13:35:26 UTC 2024 x86_64 x86_64 x86_64 GNU/Linux

Describe the bug

After trying to use the examples that use PXE and cloud-init to provision a system, none of the steps in stages: boot: were being applied. If I rename the step to boot.after: it will work.

After a lot of digging through the code, it appears that Yip detects the "#cloud-config" line at the top of the file and transpiles from cloudInit style to yipYaml which tries to merge the stages: sections for boot. I'm not sure if this is a problem when the PXE install happens or after the installation occurs.

To Reproduce

1. Create cloud-config file with "#cloud-config" at the top, auto install indicated, and "stages: boot:" section that does something that can be observed (e.g., touch a file in an ephemeral, per-boot directory)
2. PXE boot client with config_url= set to that cloud-config

Expected behavior

Installation should occur, reboot, and then apply steps from "stages: boot:"

Logs

Additional context

When the steps are renamed to "boot.after" this appears to work.

[Itxaka]

umm, I tested this with the following cc:

#cloud-config

install:
    grub_options:
        extra_cmdline: rd.immucore.debug
    poweroff: false
    reboot: false
stages:
    boot:
        - files:
            - content: "test\n"
              group: 1000
              owner: 1000
              path: /run/bar
              permissions: 511
          name: BOOT STAGE TEST
    initramfs:
        - name: Set user and password
          users:
            kairos:
                passwd: kairos
        - hostname: kairos-{{ trunc 4 .Random }}

Installed, rebooted and I can see the file there:

kairos@kairos-xhwk:~$ stat /run/bar 
  File: /run/bar
  Size: 25        	Blocks: 8          IO Block: 4096   regular file
Device: 0,26	Inode: 2335        Links: 1
Access: (0777/-rwxrwxrwx)  Uid: ( 1000/  ubuntu)   Gid: ( 1000/  ubuntu)
Access: 2024-10-02 10:11:38.036511722 +0000
Modify: 2024-10-02 10:11:38.036511722 +0000
Change: 2024-10-02 10:11:38.036511722 +0000
 Birth: 2024-10-02 10:11:38.036511722 +0000
kairos@kairos-xhwk:~$ cat /run/bar 
test                   

And I can see the stage being run in the service:

Oct 02 10:11:38 kairos-xhwk kairos-agent[1695]: 2024-10-02T10:11:38Z INF Processing stage step ''. ( commands: 0, files: 0, ... )
Oct 02 10:11:38 kairos-xhwk kairos-agent[1695]: 2024-10-02T10:11:38Z INF Processing stage step 'BOOT STAGE TEST'. ( commands: 0, files: 1, ... )
Oct 02 10:11:38 kairos-xhwk kairos-agent[1695]: 2024-10-02T10:11:38Z INF Done executing stage 'boot'

And this is with an example from the examples docs you linked. The only things changed were to add a name to the stage so I could see it on the logs and setting the group/owner to 1000 so it picks up the default ubuntu user on my iso.

If you want us to solve this we will need:

• full cc you used
• logs from the boot stage service (journalctl -u cos-setup-boot)

Also can you confirm that this happens AFTER install, i.e. from an installed system? During installation (livecd/pxe) im not sure its possible for the boot service to read from the config_url as that is ingested after it has run, so it would be good to clear that the reproduction is as follow:

• Cloud Config has some stages.boot actions (create file, set hostname, whatever)
• Boot from IPXE with config_url
• Install
• Reboot to installed system
• Boot service runs but the boot stages are not executed
• Moving those to boot.after makes them run

Thanks!

[jimmykarily]

I'm closing this since because it's stalled. Feel free to re-open if you have more information to help us reproduce.