Adapting kubespray to deploy k8s within Embassy cloud portal

• get latest kubespray
• deploy
• start k8s-galaxy
• trigger k8s-galaxy workflow execution

after checkout and getting the latest kubespray, recreating some symlinks might be necessary:

• .//extra_playbooks/roles -> ../roles
• .//extra_playbooks/inventory -> ../inventory
• .//contrib/terraform/group_vars -> ../../inventory/group_vars
• .//contrib/terraform/openstack/hosts -> ../terraform.py
• .//contrib/terraform/openstack/group_vars -> ../../../inventory/group_vars
• .//contrib/network-storage/glusterfs/group_vars -> ../../../inventory/group_vars
• .//contrib/network-storage/glusterfs/roles/bootstrap-os -> ../../../../roles/bootstrap-os
• ./kubespray ->

troubleshooting

in case of ssh connection issues check if there are still ssh folders in /tmp lingering

rm /tmp/kubespraytest-k8s-*

also cleanse known_hosts from bastion public_ip

to bring up the deployment

ostack/deploy.sh

before that, except for sourcing the openstack credentials, the following exports might be needed:

set environment variables used by scripts in cloud-deploy/

export PUBLIC_KEY="" export PRIVATE_KEY="" export PORTAL_DEPLOYMENTS_ROOT="$PWD/deployments" export PORTAL_APP_REPO_FOLDER="$PWD" export PORTAL_DEPLOYMENT_REFERENCE="deployment-ref-ubuntu"

deployment_dir="$PORTAL_DEPLOYMENTS_ROOT/$PORTAL_DEPLOYMENT_REFERENCE" if [[ ! -d "$deployment_dir" ]]; then mkdir -p "$deployment_dir" fi printf 'Using deployment directory "%s"\n' "$deployment_dir"

export TF_VAR_kube_api_pwd="" export TF_VAR_cluster_name="" export TF_VAR_number_of_etcd="0" export TF_VAR_number_of_k8s_masters="1" export TF_VAR_number_of_k8s_masters_no_floating_ip="0" export TF_VAR_number_of_k8s_masters_no_etcd="0" export TF_VAR_number_of_k8s_masters_no_floating_ip_no_etcd="0" export TF_VAR_number_of_k8s_nodes_no_floating_ip="2" export TF_VAR_number_of_k8s_nodes="0" export TF_VAR_public_key_path="" export TF_VAR_image="Ubuntu16.04" #export TF_VAR_image="CentOS7-1612" export TF_VAR_ssh_user="ubuntu" #export TF_VAR_ssh_user="centos" #export TF_VAR_flavor_k8s_node="e9ca7478-7957-4237-b3d0-d4767e1de65f" #ext5 export TF_VAR_flavor_k8s_node="11" #export TF_VAR_flavor_etcd="12" #export TF_VAR_flavor_k8s_master="91ba172b-cb4c-453c-b7fc-56cb79c78968" #ext5 export TF_VAR_flavor_k8s_master="11" export TF_VAR_network_name="" export TF_VAR_floatingip_pool=""

GlusterFS variables

#export TF_VAR_flavor_gfs_node="e9ca7478-7957-4237-b3d0-d4767e1de65f" #ext5 export TF_VAR_flavor_gfs_node="11" export TF_VAR_image_gfs="Ubuntu16.04" export TF_VAR_number_of_gfs_nodes_no_floating_ip="3" export TF_VAR_gfs_volume_size_in_gb="30" export TF_VAR_ssh_user_gfs="ubuntu"

export KUBELET_DEPLOYMENT_TYPE="host" export KUBE_VERSION="v1.8.2"

to destroy the deployment

use

terraform destroy --force --input=false --state=deployments/deployment-ref-ubuntu/terraform.tfstate