Debian 11 boots into Debian 10 Kernel in VTE

In the VTE when creating an experiment with the debian:11 image, the node boots into the debian:10 kernel.

The sledctl output for the node looks like this, which seems to suggest that I should be landing in a Debian 11 kernel.

rvn@cmdr:~$ sledctl --server 172.31.0.2 get command 04:70:00:01:70:1e
mac:	04:70:00:01:70:1e
id:	a318a7e0
time:	Sun, 26 Apr 2020 14:44:41 UTC
wipe: nil
write: 	complete
	time:		Sun, 26 Apr 2020 14:45:03 UTC
	image:		debian-11-disk
	kernel:		debian-11-kernel
	initrd:		debian-11-initramfs
	device:		sda
kexec: 	in progress
	time:		Sun, 26 Apr 2020 14:45:03 UTC
	append:		root=PARTUUID=a0000000-0000-0000-0000-00000000000a rootfstype=ext4 rw net.ifnames=0 biosdevname=0

However when grabbing a console onto the node I see this

image

A quick look at a storage node shows distinct file sizes for Debian 10 and 11 kernels

rvn@stor0:/var/img$ ls -l deb*
rw-r--r-- 1 root root 2147483648 Apr 25 19:48 debian-10-disk
rw-r--r-- 1 root root   25810604 Apr 25 19:48 debian-10-initramfs
rw-r--r-- 1 root root    5270768 Apr 25 19:48 debian-10-kernel
rw-r--r-- 1 root root 2147483648 Apr 25 19:48 debian-11-disk
rw-r--r-- 1 root root   26470573 Apr 25 19:48 debian-11-initramfs
rw-r--r-- 1 root root    5627632 Apr 25 19:48 debian-11-kernel

VTE version info

$ git rev-parse --short HEAD
 1ed2cef

$ git submodule status
 ba55ab35ea6c0b1a93d0508c9d8637bef87446f1 experiments (heads/master)
 6d713adb7818b2f88208e95deeb22c87f3395d45 tb-deploy (heads/master)

Rebooting the machine does land me in the correct kernel.

This is a known bug for the vte with sled v0.7.0.

Fix:

Run this ansible-playbook.

- hosts: [stor0,stor1,stor2,stor3]
  become: true
  vars:
    - image_directory: "/var/img/pxe"
    - tag: testing
    - ci_path: https://gitlab.com/mergetb/tech/sled/-/jobs/artifacts
    - timeout: 120
  tasks:
    - name: download all initramfs and kernels (rohu/ahi, minnow)
      get_url:
        url: "{{ ci_path }}/{{ tag }}/raw/build/{{ item.name }}?job=make"
        dest: "{{ image_directory }}/{{ item.dest }}"
        force: yes
        backup: yes
        timeout: "{{ timeout }}"
      loop:
        - {"name": "generic-bzImage", "dest":"sledc-kernel"}
        - {"name": "generic-dhcp-initramfs.cpio", "dest":"sledc-dhcp-initramfs"}

This changes the sledc image from v0.7.0 to testing which fixes the write issues.

This issue has a temporary workaround in place by having the VTE specify the sledTag=testing overriding the tb-deploy default of v0.7.0.