Kubelet install failure and deprecation warning fix (#58)

[chris-short]

• Squashed commit of the following:

commit 22cad03f2d0fe03da1abed89a3f11e5cf684b0b5 Author: Chris Short chris@chrisshort.net Date: Mon Jan 15 17:52:36 2018 -0500

removing ignorable files

commit 89e1199ccbd24ba495789b79fcffcdae90010f57 Author: Chris Short chris@chrisshort.net Date: Mon Jan 15 17:47:06 2018 -0500

Updating .gitignore

Signed-off-by: Chris Short <chris@chrisshort.net>

commit e90c58b8e785aebc4432b53fe08935d3f16012ad Author: Chris Short chris@chrisshort.net Date: Mon Jan 15 17:43:53 2018 -0500

Initial rak8s Commit

Signed-off-by: Chris Short chris@chrisshort.net

• Set theme jekyll-theme-tactile

• Create CNAME

• link to how to change size of video memory

Closes #1.

• Update mechanism to enable cgroup memory

See: https://github.com/raspberrypi/linux/commit/ba742b52e5099b3ed964e78f227dc96460b5cdc0

• why not both?

I found the elimination of cgroup_enable=memory actually broke my raspbian kubernetes setup. putting both worked with my testing after reboot.

• k8s upgrade role init

Signed-off-by: Chris Short chris@chrisshort.net

• upgrade playbook ready

Signed-off-by: Chris Short chris@chrisshort.net

• Updating README

Adding link to Discourse site

Signed-off-by: Chris Short chris@chrisshort.net

• Adding Code of Conduct

Long overdue

Signed-off-by: Chris Short chris@chrisshort.net

• Create issue_template.md

• issue_template.md: release file path correction

/etc/os-release has the preferred data

Signed-off-by: Chris Short chris@chrisshort.net

• Changed deprecated URL https://git.io/weave-kube to https://cloud.weave.works/k8s/net?k8s-version=

• Adding PR template

• Move TASK [Pass bridged IPv4 traffic to iptables' chains] after TASK [kubeadm : Run Docker Install Script]

• Changed networking from weave to flannel. Added version control for docker-ce, kubeadm, kubelet, kubectl, kubernetes images and flannel.

• Added cleanup playbook that removes kubernetes, all images, docker-ce, kubeadm, kubelet, kubectl and performs a reboot.

• Cleaning Up/Organizing/Adding Functionality

• Incorporate changes from #34

• Changing soon to be deprecated Ansible syntaxes

• Adding a backup option for /root/.kube/config

• Set hostname from inventory

• Adding version options to group_vars

Signed-off-by: Chris Short chris@chrisshort.net

• Backup Option

Signed-off-by: Chris Short chris@chrisshort.net

• Add troubleshooting steps and timing adjustments (#37)

• Troubleshooting Steps

updating README with some info that helped me when following the documentation

• Timeout Changes

Allows for slowed nodes to be recognized

• Timeout Changes

• v0.2.1

This version matching appears to be working as desired on three Raspberry Pi 3B+

The inventory file is breaking up my six node stack into two clusters. This will enable easier development.

A force flag has been added to kubeadm reset

Signed-off-by: Chris Short chris@chrisshort.net

• Fix 'Need to Start Over' script name (#40)

• Applying podnet variable correctly

Not sure how that even happened.

Signed-off-by: Chris Short chris@chrisshort.net

• Lock docker version to currently installed version when it is installed (#44)

• Ignore failing deletion of swap (#46)

• Due to using hypriot, which doesn't have dphys-swapfile installed by default, the current deletion process fails. Adding ignore_errors fixes this.

TASK [kubeadm : Disable Swap] ** fatal: [rak8s000]: FAILED! => {"changed": true, "cmd": "dphys-swapfile swapoff && dphys-swapfile uninstall && update-rc.d dphys-swapfile remove", "delta": "0:00:00.005302", "end": "2018-12-24 20:35:44.430288", "msg": "non-zero return code", "rc": 127, "start": "2018-12-24 20:35:44.424986", "stderr": "/bin/sh: 1: dphys-swapfile: not found", "stderr_lines": ["/bin/sh: 1: dphys-swapfile: not found"], "stdout": "", "stdout_lines": []} ...ignoring

• reduce gpu memory (#48)

• update dashboard url (#47)

• K8S patch and other clean up items (#51)

• updated ansible configs to no longer use control paths as this failed on ubuntu 16.04 when deploying

• 50 - updated k8s version to latest release, updated docker, using flannel master currently as this is the only version that worked out of the box with k8s 1.13.1 - jh

• moved python apt dependency below apt-get update to allow for new installations to update cache prior installing, fixed reboot with logic from cleanup.yml

• removed apt-get update from kubeadm main since this is in common main and added environment var to stop apt from being angry - jh

• flannel at master now supports arm and amd64 k8s labels, as such we don't need to deal with this anymore.

• more updates for 1.14, removed gpu check since it's not longer working on my pi's and failing with a nonzero exit code.

• Altered apt syntax

Packages specified as an array in pkg rather than in with_items resolves "Invoking 'apt' only once while using a loop via squash_actions is deprecated" warning

• update_cache added

Resolves "No package matching 'kubelet' is available" error

Description

Testing

Issue Number

My change fixes issue #

[chris-short]

@PostlMC I don't think this is what you wanted.

[PostlMC]

Ugh, yeah. Something's not right here. I'm not sure what I did that made my PR want to roll up all this.

[exotime]

For what it's worth, this branch worked for me when master did not.

[MatthiasLohr]

Anything new on this?

[chris-short]

See Issue #60 for resolution.