This Openstack Heat Template creates up a Kubernetes cluster on Openstack for Data Together resources. A HEAT Orchestration Template are used for both provisioning and configuration, however, this can be modified, to allow for tools like Ainsible to perform the configuration and/or the provisioning as well.
New features of the Data Together platform are coordinated and developed in one or more of the other repositories listed in our Roadmap.
Key pairs are ssh credentials which are injected into images when they are launched. Creating a new key pair registers the public key and downloads the private key (a .pem file). Protect and use the key as you would any normal ssh private key.
-
Navigate to Compute > Access & Security > Key Pairs > Create Key Pair
-
Provide a name for the key and download and save the .pem file to the .ssh folder on your local machine. Change the permission of the file to 600. (
chmod 600 ____.pem)
- Openstack Auth URL : Obtain from Compute > Access & Security > API Access > Download Openstack RC
- Tenant ID : Obtain from Compute > Access & Security > API Access > Download Openstack RC
- External Network : See Routers tab for name of the External Network.
-
Navigate to Orchestration > Stacks > Launch Stack
-
Select URL as the Template Source, and provide the RAW URL of the heat template. Alternatively, you may copy/paste the contents or upload the file. URL:
-
Fill out Launch Stack form:
- Use the Openstack Auth URL, Tenant ID, External Network and Key Pair from above.
- You have the option to Use Kubeadm to bootstrap the deployment. If used, the latest version of Kubernetes is installed. If it is not used, the latest stable version of the CentOS repository Kubernetes (current at 1.5) is installed.
- You may provide an additional url to a file to run Custom Scripts on the master node if needed.
- The selected Image should be
Centos 7 x64-2017-02. [See Customisation section for details] - The Flavor selected should ideally at least 2 processors with 6 GB of RAM.
- You can also choose the number of Minions and the Minion Volume Size. The Master node's volume is 8GB by default.
- You will know the system is ready when the k8s_master log shows
"**** COMPLETE ****". - Log in using this command:
ssh -i .ssh/<ssh_key_name>.pem centos@<k8s_masterip>. Outputs in the Stack Overview for Master node log in instructions post-deployment. (Orchestration > Stakcs > (Select your stack) > Overview)
Checks & Troubloshooting:
| Test | Expected Outcome | Troubleshoot |
|---|---|---|
kubectl get nodes --all-namespaces=true |
Should list all nodes in Ready state. | If all nodes are visible and in Not Ready, check whether .kube/config has the same contents as /etc/kubernetes/admin.conf, and whether the networking addon (weave or flannel) is properly set up. |
| If one or more nodes are missing, check the logs of the missing nodes through the Openstack dashboard. | ||
systemctl is-system-running |
running |
run systemctl --state=failed to determine failed services |
| - | Issues | Troubleshoot |
| - | vm set up stalled | Delete stack. Try rebooting the Network (Network > Networks > Edit Network. Set Admin State as DOWN. Refresh and set again as UP.) |
The nodes have a GO environemnt installed, but the following are optional installation guides:
A tool for managing Kubernetes charts. Charts are packages of pre-configured Kubernetes resources.
curl -s https://raw.githubusercontent.com/helm/helm/master/scripts/get > get_helm.sh
chmod 700 get_helm.sh
./get_helm.sh
kubectl create serviceaccount --namespace kube-system tiller
kubectl create clusterrolebinding tiller-cluster-rule --clusterrole=cluster-admin --serviceaccount=kube-system:tiller
kubectl patch deploy --namespace kube-system tiller-deploy -p '{"spec":{"template":{"spec":{"serviceAccount":"tiller"}}}}'
helm init --service-account tiller --upgrade
helm repo update
Reference:
A convenient way to unlock the power of dynamically provisioned, persistent GlusterFS volumes in Kubernetes.
yum -y -q install centos-release-gluster41.x86_64 glusterfs gluster-cli glusterfs-libs glusterfs-server
systemctl enable glusterd.service
systemctl start glusterd.service
Reference: *GlusterFS
Resource management and statistics
kubectl apply -f https://raw.githubusercontent.com/kubernetes/dashboard/master/src/deploy/recommended/kubernetes-dashboard.yaml
kubectl apply -f https://github.com/giantswarm/kubernetes-prometheus/tree/master/manifests/grafana/grafana-import-dashboards-job.yaml
Reference:
For Logs and Notes: See DataTogether Shared Drive.
Heat template Version: 2015-04-30
Policies and IPs
k8s_master_floating- Creates a floating IP address for the master nodek8s_master_floating_association- Attaches it to the master nodek8s_master_access- Creates a security group for the master node with the necessary open ports
Servers and Volumes
k8s_master- Creates the master nodek8s_master_vol- Creates the master volumek8s_master_vol_attach- Attaches the volume to the master nodek8s_minion_group- Creates a group of minions with volumesk8s_minion- Creates a minion nodek8s_mimnion_vol- Creates the minion node volumesk8s_minion_vol_attach- Attaches the volume to the minion node
Outputs
ssh_key_name- key pair used for remote accessk8s_masterip- External IP of the master nodelogin_instructions- Guideline on how to log into the master node
If switching to a different image, change the references to centos to the default login for that image, and adjust yum to the default package installer. See the functions os_inits, packages_install and packages_install_ka which are image specific in both the master and minion templates, and kubeadm_script in the minion template. Since the OS optimation is completely in shell scripting, the template can be reworked for use with other deployment tools like Ainsible for other platforms.
Copyright (C) 2018 Data Together
See the LICENSE file for details.
We would love involvement from more people! If you notice any errors or would like to submit changes, please see our Contributing Guidelines.
We use GitHub issues for tracking bugs and feature requests and Pull Requests (PRs) for submitting changes.