On CoreOS, without this, it fails to kubectl apply MetalLB due to lack of privileges.

Pods on down/unresponsive nodes can't be deleted without
--force --grace-period=0.

Fixes #4314

* Move most coredns templates to static files

This should speed up the task slightly

* yaml lint fixes

hosts are ordered dictionary
remove ansible_user from inventory file

- fix order of entries when the new yaml file is created
- fix group in case there are no hosts in it

* add blockSize to IPPool spec for Calico >= v3.3.0

* fix "cidr" spec in Calico IPPool resource for my PR

* Remove hard dependence on facts for all nodes

* Update main.yaml

* Update main.yaml

* rewrite inventory script to create inventory file in YAML format

* minor fixes to inventory script

* change requirments for the inventory script

* Add 1.12.6, sort arm64 descending

* remove 1.10.x checksums (EOL anyways)

* feat(external-provisioner/local-path-provisioner): adds support for local path provisioner

Helpful for local development but also in production workloads (once the
permission model is worked out) where you have redundancy built into the
software uses the PVCs (e.g. database cluster with synchronous
replication)

* feat(local-path-provisioner): adds debug flag, image tag group var

* fix(local-path-provisioner): moves image repo/tag to download role

* test(gce_centos7-flannel): enables local-path-provisioner in test case

* fix(addons): add image repo/tag to commented default values

* fix(local-path-provisioner): typo in jinja template for local path provisioner

* style(local-path-provisioner): debug flag condition re-formatted

* fix(local-path-provisioner): adds missing default value for debug flag

* fix(local-path-provisioner): syntax fix for debug if condition end

* fix(local-path-provisioner): jinja template syntax: if condition white space

kube ansible module can be used with recursive: true
which sill process the directory used in -f, --filename recursively

The openstackclient is the preferred CLI for OpenStack
environments and should be used instead of novaclient.

* Create packet.md

* Update README.md

* Update README.md

* Update packet.md

download the latest version

* Update packet.md

* 30MiB for gpu-device-plugin

* use vars for easier configuration

Upstream Changes:

  - weave 2.5.1 (https://github.com/weaveworks/weave/releases/tag/v2.5.1)

Our Changes:

  - Sync templates with upstream changes

* added hardware requirements in README.md

* added hardware requirements in README.md

* fact gathering should run only once per node
* eliminate ansible version check, it is at the beginning of each
  playbook

The lastest version of ubuntu-nvidia-driver-installer contains a fix for
https://github.com/GoogleCloudPlatform/container-engine-accelerators/issues/90
which causes the installer pod to crash when driver is already loaded.

This was already approved in #4106 but there are CI issues
with that PR due to references to kubernetes incubator.

After upgrading to Kubespray 2.8.1 with Kubeadm enabled Rook
Ceph volume provision failed due to the flexvolume plugin dir not
being correct. Adding the var fixed the issue

Useful if the default 20GB is not enough in cases where you are using
the local path provisioner of rancher for example