Rancher Installation
The following are instructions on how to create an Openstack VM running Rancher.
Launch new VM instance to host the Rancher Server
Select Ubuntu 16.04 as base image
Select "No" on "Create New Volume"
Select Flavor
Known issues exist if flavor is too small for Rancher. Please select a flavor with at least 4 vCPU and 8GB ram.
Networking
Security Groups
Key Pair
Use an existing key pair (e.g. onap_key), import an existing one or create a new one to assign.
Apply customization script for the Rancher VM
This customization script will:
- setup root access to the VM (comment out if you wish to disable this capability and restrict access to ssh access only)
- install docker *
- install rancher *
- install kubectl *
- install helm * 2.8.2
- install nfs server
* ONAP release supported version (Helm 2.8.2 not 2.7.2 in the screen cap below as of May 2018
Launch Instance
Assign Floating IP for external access
Kubernetes Installation
Launch new VM instance(s) to create a Kubernetes single host or cluster
To create a cluster:
- do not append a '-1' suffix (e.g. sb4-k8s)
- increase count to the # of of kubernetes worker nodes you want (eg. 3)
Select Ubuntu 16.04 as base image
Select "No" on "Create New Volume"
Select Flavor
The size of a Kubernetes host depends on the size of the ONAP deployment that will be installed.
As of the Beijing release a minimum of 3 x 32GB hosts will be needed to run a full ONAP deployment (all components).
If a small subset of ONAP components are being deployed for testing purposes, then a single 16GB or 32GB host should suffice.
Networking
Security Group
Key Pair
Use an existing key pair (e.g. onap_key), import an existing one or create a new one to assign.
Apply customization script for Kubernetes VM(s)
This customization script will:
- setup root access to the VM (comment out if you wish to disable this capability and restrict access to ssh access only)
- install docker *
- install kubectl *
- install helm *
- install nfs common (see configuration step here)
* ONAP release supported version
Launch Instance
Assign Floating IP for external access
Setting up an NFS share for Multinode Kubernetes Clusters
The figure below illustrates a possible topology of a multinode Kubernetes cluster.
One node, the Master Node, runs Rancher and Helm clients and connects to all the Kubernetes nodes in the cluster. Kubernetes nodes, in turn, run Rancher, Kubernetes and Tiller (Helm) agents, which receive, execute, and respond to commands issued by the Master Node (e.g. kubectl or helm operations). Note that the Master Node can be either a remote machine that the user can log in to or a local machine (e.g. laptop, desktop) that has access to the Kubernetes cluster.
Deploying applications to a Kubernetes cluster requires Kubernetes nodes to share a common, distributed filesystem. One node in the cluster plays the role of NFS Master (not to confuse with the Master Node that runs Rancher and Helm clients, which is located outside the cluster), while all the other cluster nodes play the role of NFS slaves. In the figure above, the left-most cluster node plays the role of NFS Master (indicated by the crown symbol). To properly set up an NFS share on Master and Slave nodes, the user can run the scripts below.
The master_nfs_node.sh script runs in the NFS Master node and needs the list of NFS Slave nodes as input, e.g.:
sudo ./master_nfs_node.sh node1_ip node2_ip ... nodeN_ip
The slave_nfs_node.sh script runs in each NFS Slave node and needs the IP of the NFS Master node as input, e.g.:
sudo ./slave_nfs_node.sh master_node_ip
Configuration (Rancher and Kubernetes)
Access Rancher server via web browser
(e.g. http://10.12.6.16:8080/env/1a5/apps/stacks)
Add Kubernetes Environment to Rancher
1. Select “Manage Environments”
2. Select “Add Environment”
3. Add unique name for your new Rancher environment
4. Select the Kubernetes template
5. Click "create"
6. Select the new named environment (ie. SB4) from the dropdown list (top left).
Rancher is now waiting for a Kubernetes Host to be added.
Add Kubernetes Host
1. If this is the first (or only) host being added - click on the "Add a host" link
and click on "Save" (accept defaults).
otherwise select INFRASTRUCTURE→ Hosts and click on "Add Host"
2. Enter the management IP for the k8s VM (e.g. 10.0.0.4) that was just created.
3. Click on “Copy to Clipboard” button
4. Click on “Close” button
Without the 10.0.0.4 IP - the CATTLE_AGENT will be derived on the host - but it may not be a routable IP.
Configure Kubernetes Host
1. Login to the new Kubernetes Host
2. Paste Clipboard content and hit enter to install Rancher Agent
Return to Rancher environment (e.g. SB4) and wait for services to complete (~ 10-15 mins)
Configure kubectl and helm
Note that in this example we are configuring kubectl and helm that have been installed (as a convience) onto the rancher and kubernetes hosts.
Typically you would install them both on your PC and remotely connect to the cluster. The following procedure would remain the same.
1. Click on CLI and then click on “Generate Config”
2. Click on “Copy to Clipboard” - wait until you see a "token" - do not copy user+password - the server is not ready at that point
3. Create a .kube directory in user directory (if one does not exist)
4. Paste contents of Clipboard into a file called “config” and save file
4. Validate that kubectl is able to connect to the kubernetes cluster
and show running pods
5. Validate helm is running at the right version.
If not, an error like this will be displayed:
6. Upgrade the server-side component of helm (tiller) via ‘helm init --upgrade’
ONAP Deployment via OOM
Now that kubernetes and Helm are installed and configured you can prepare to deploy ONAP.
Until an LF-hosted public ONAP repository is available (comping soon!), please clone the OOM repo (https://gerrit.onap.org/r/gitweb?p=oom.git;a=summary).
Follow the instructions in the oom/kubernetes/README.MD or look at the official documentation to get started
http://onap.readthedocs.io/en/latest/submodules/oom.git/docs/oom_user_guide.html