Kubeadm download fails on armv7l

[nmiculinic]

Environment:

• Cloud provider or hardware configuration:
  Hybrid. Masters are on AWS, workers are on Beablebone Black

• OS (printf "$(uname -srm)\n$(cat /etc/os-release)\n"):
  debian@bbb-test:~$ printf "$(uname -srm)\n$(cat /etc/os-release)\n"
  Linux 4.19.9-ti-r5 armv7l
  PRETTY_NAME="Debian GNU/Linux 9 (stretch)"
  NAME="Debian GNU/Linux"
  VERSION_ID="9"
  VERSION="9 (stretch)"
  ID=debian
  HOME_URL="https://www.debian.org/"
  SUPPORT_URL="https://www.debian.org/support"
  BUG_REPORT_URL="https://bugs.debian.org/"

• Version of Ansible (ansible --version):
  ansible 2.7.1
  config file = /Users/lpp/Desktop/ascalia/bb/ansible.cfg
  configured module search path = [u'/Users/lpp/.ansible/plugins/modules', u'/usr/share/ansible/plugins/modules']
  ansible python module location = /usr/local/lib/python2.7/site-packages/ansible
  executable location = /usr/local/bin/ansible
  python version = 2.7.15 (default, Oct 2 2018, 11:47:18) [GCC 4.2.1 Compatible Apple LLVM 10.0.0 (clang-1000.11.45.2)]

Kubespray version (commit) (git rev-parse --short HEAD):
bb (master) ✗ git rev-parse --short HEAD k8s [kubernetes-admin@bbb.local]
Alias tip: g rev-parse --short HEAD
c204794

Network plugin used:
weave

Ansible failed output:

fatal: [bbb-test]: FAILED! => {"attempts": 4, "changed": false, "dest": "/tmp/releases/kubeadm", "msg": "Request failed: <urlopen error no host given>", "state": "absent", "url": "https://storage.googleapis.com/kubernetes-release/release/v1.13.1/bin/linux/armv7l\n/kubeadm"}

EDIT: Closed in favor of #4294

[laimison]

any luck on this issue?

[nmiculinic]

Hmm... I figured the arch was wrongly detected since it was armhf. In the end, I gave up on kubeadm (( you could download it from debian repos for armhf )) since there were some issues during installation.

I did it by hand on the arm, setup kubelet manually with systemd and it kinda works. In the process, I learned kubelet is pretty resource intensive for such weak hardware. Further research pending.

[laimison]

Thanks for your reply. Are you using Raspberry Pi? I'm wondered if you can install armhf images to this?

My one is Raspberry Pi 3 B+ and the same architecture detected in /etc/*release which is armv7l.

I'm stuck at similar error, but different image:

FAILED - RETRYING: container_download | Download containers if pull is required or told to always pull (all nodes) (2 retries left).
fatal: [raspberrypi-master-1.lan]: FAILED! => {"attempts": 4, "changed": true, "cmd": ["/usr/bin/docker", "pull", "gcr.io/google_containers/cluster-proportional-autoscaler-armv7l", ":1.3.0"], "delta": "0:00:00.301587", "end": "2019-02-02 01:08:19.666244", "msg": "non-zero return code", "rc": 1, "start": "2019-02-02 01:08:19.364657", "stderr": "\"docker pull\" requires exactly 1 argument.\nSee 'docker pull --help'.\n\nUsage:  docker pull [OPTIONS] NAME[:TAG|@DIGEST]\n\nPull an image or a repository from a registry", "stderr_lines": ["\"docker pull\" requires exactly 1 argument.", "See 'docker pull --help'.", "", "Usage:  docker pull [OPTIONS] NAME[:TAG|@DIGEST]", "", "Pull an image or a repository from a registry"], "stdout": "", "stdout_lines": []}

Why you didn't try to replace gcr.io/google_containers to another repo in kubespray (for example to Debian)?
If there are cluster-proportional-autoscaler-arm* and kubeadm packages available.

[laimison]

I'm wondered that I'm not able to find any results on Google search by using cluster-proportional-autoscaler-armv7l or cluster-proportional-autoscaler-armhf and kubespray by default points to gcr.io/google_containers/cluster-proportional-autoscaler-armv7l which is not available at all. Any hints on this? Is that the reason why you ended up by configuring manually?

[b23prodtm]

Turns out that k8s-dns-dnsmasq-nanny-arm is in kubernetes version 1.14.4 which fixed that issue
kubernetes/kubernetes#48090

[lwolf]

had the same issue.
ansible_architecture_rename rule does not work as expected on armv7, it sets host_architecture value as armv7l\n

Reformatting the rule es shown below, helps.

- name: ansible_architecture_rename
  set_fact:
    host_architecture: >-
      {%- if ansible_architecture in architecture_groups -%}
      {{architecture_groups[ansible_architecture]}}
      {%- else -%}
       {{ansible_architecture}}
      {% endif %}

- name: ansible_architecture_rename
  set_fact:
    host_architecture: "{%- if ansible_architecture in architecture_groups -%}{{architecture_groups[ansible_architecture]}}{%- else -%}{{ansible_architecture}}{% endif %}"

[b23prodtm]

Found this helps 👍

- set_fact:
    architecture_groups:
      x86_64: amd64
      aarch64: arm64
      armv7l: arm

[nmiculinic]

How about we close this issue in favor of #4294 ?