/run/cos/active doesn't correctly size based on the ISO

[cog-nick]

Kairos version:

PRETTY_NAME="Ubuntu 22.04.3 LTS"
NAME="Ubuntu"
VERSION_ID="22.04"
VERSION="22.04.3 LTS (Jammy Jellyfish)"
VERSION_CODENAME=jammy
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=jammy

CPU architecture, OS, and Version:

Linux controller 5.19.0-46-generic #47~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Wed Jun 21 15:35:31 UTC 2 x86_64 x86_64 x86_64 GNU/Linux

Describe the bug

When I use a custom base Docker Image (Ubuntu + Packages), /run/cos/active runs out of size (the loopback) on the initial install. This is likely due to it not being sized correctly w.r.t to the ISO

To Reproduce

Docker Image:

FROM quay.io/kairos/kairos-ubuntu-22-lts:v2.3.1-k3sv1.26.6-k3s1
RUN add-apt-repository ppa:rmescandon/yq && apt-get update && apt-get upgrade -y && apt-get install -y zip unzip fdisk rsyslog python2 python2-dev yq dkms lsscsi git build-essential ntp ntpstat htop tree autossh libusb-1.0-0-dev libpcap-dev libaudit-dev libcupti-dev bridge-utils lm-sensors ifenslave iotop traceroute smartmontools nmap nut nut-client nut-server \
  && apt-get clean && rm -rf /var/lib/apt/lists/*
# Add latest Kairos Agent
ADD kairos-agent /usr/bin/kairos-agent
# Add helm charts
COPY ingress-nginx*.tgz /var/lib/rancher/k3s/server/static/charts/
COPY cert-manager*.tgz /var/lib/rancher/k3s/server/static/charts/
COPY rancher-*.tgz /var/lib/rancher/k3s/server/static/charts/
COPY harbor-*.tgz /var/lib/rancher/k3s/server/static/charts/
COPY rabbitmq-*.tgz /var/lib/rancher/k3s/server/static/charts/
COPY kyverno-*.tgz /var/lib/rancher/k3s/server/static/charts/

I think you'll see the same issue without the Helm charts since until adding packages from apt, packaging helm charts for air gap install worked without a hitch.

Cloud Init:

#cloud-config
​
users:
- name: "user"
  passwd: "user"
  lock_passwd: false
  no-create-home: false
  shell: /bin/bash
  homedir: "/home/user"
  groups: ["admin"]
  ssh_authorized_keys: []
​
install:
  device: "/dev/sda"
  reboot: true
  auto: true
  partitions:
    oem:
      size: 4096
      fs: ext4
    state:
      size: 20480
      fs: ext4
    recovery:
      size: 12288
      fs: ext4
  system:
    size: 8192
  passive:
    size: 8192
  recovery-system:
    size: 8192

Expected behavior

That it successfully installs by correct sizing /run/cos/active to be larger than a 3.0GB loopback device.

Logs

I can't capture it when from the target device, but it's your standard "out of space" error when you fill a disk on Linux.

Additional context

[Itxaka]

Looks like its ignoring the size set on the config file:

[Itxaka]

this is kairos 2.3.1 which ships agent v2.1.12 which doesnt have a lot of the improvements required for the auto-size or override of system size via cloud-config values.

talked to @cog-nick in slack and recommended bumping the agent to latest stable version which should fix this.

If this doesnt work or bumping the agent doesnt work for you we can still workaround this.

The old method was to have a /etc/elemental/config.yaml file with those values, which kairos-agent read and use for setting several values. We can override those values by creating a file under that path with the following values:

/etc/elemental/config.yaml

install:
  system:
    size: 8192
  passive:
    size: 8192
  recovery-system:
    size: 8192

Notice that this file does not need the #cloud-config stanza at the top, this is a legacy way of setting things based of elemental-cli: https://github.com/rancher/elemental-cli/blob/main/config.yaml.example

[Itxaka]

This seems like it would be fixed based on the feedback and improvements to the size calculation. Feel free to reopen if needed.