This README will guide you to installing Openshift 3.x on vSphere using Terraform. Centos 7 is the only OS supported by the install documented here.
Note: There is no hard requirement to use Terraform, or to Install on vSphere. All that is required are the correct number of virtual machines and the required DNS records are in place.
Configure External DNS with the following records
*.app.lab.unixcraft.lcl A Record <IP address of Infra Node>
osmaster.lab.unixcraft.lcl A Record <IP address of Master Node>
osinfra.lab.unixcraft.lcl A Record <IP address of Master Node>
osnode01.lab.unixcraft.lcl A Record <IP address of Master Node>
osnode02.lab.unixcraft.lcl A Record <IP address of Master Node>
osnode03.lab.unixcraft.lcl A Record <IP address of Master Node>
Note: The domain here is used as an example
Terraform can be used to create the required virtual machines (Please note - the terraform files will need to be edited with the correct IP addresses and domain names). if you do not want to use Terraform or vSphere, here are the details for the machines.
Note: Terraform, does not seem to add disks in ordered fashion, so to get around this issue, comment out the GFS drives in the osnode section and run the terraform apply. Once completed, uncomment the GFS drives and re-apply. This will ensure the drives come as sdb for Docker and sdc for GFS
Once the Terraform file has been edited, execute the following command
terraform apply --var-file=/<path to terraform variable file>/openshift_lab.tfvars
If you wish to create the VM's in some other way, here are the requirenents.
1 * Master Node - (2vCPU, 8GB memory, 30GB HDD (OS), 30GB (Docker)) 1 * Infra Node - (2vCPU, 16GB memory, 30GB HDD (OS), 30GB (Docker)) 3 * Virtual Machines - (2vCPU, 8GB memory, 30GB HDD (OS), 30GB (Docker), 50GB (GFS))
Note: All of the VM's should be on a single flat network.
All of the virtual machines need to be configured with the following
- Ansible SSH user and key
- Disable firewall and SE-linux
These steps will not be detailed here
Assuming you have Ansible installed and configured on the machine which will executes the plays. There is one pre-requisties that needs to be installed for the installation to be successful.
Install the python-passlib module.
pip install passlib
Note: If you are on a Mac running Sierra and you have used Homebrew to install Python and Ansible, please ensure you have created a symlink for python2 to python in
/usr/local/bin
Also make sure that /usr/local/bin is in your PATH. This is important as when Ansible executes it may use the system Python and will not find the passlib module and fail.
Clone the following repositories into /tmp
cd /tmp
Openshift vSphere installation repositories
git clone https://github.com/f1gjam/unixcraft-openshift.git
Official openshift-ansible Repo
git clone https://github.com/openshift/openshift-ansible
Edit the Ansible inventory file inside the unixcraft-openshift repositories
cd /tmp/unixcraft-openshift/inventories/
Edit the following:
ansible_ssh_private_key_file=<path to ssh key file to be used by ansible>
Ensure the path to the private key used by ansible for accessing the virtual machines is correct
Edit the GlusterFS section and update the IP addresses of the OSNODE's
[glusterfs]
osnode01.lab.unixcraft.lcl glusterfs_ip=192.168.10.62 glusterfs_devices='[ "/dev/sdc" ]'
osnode02.lab.unixcraft.lcl glusterfs_ip=192.168.10.63 glusterfs_devices='[ "/dev/sdc" ]'
osnode03.lab.unixcraft.lcl glusterfs_ip=192.168.10.64 glusterfs_devices='[ "/dev/sdc" ]'
Once the inventory file has been updated you can execute the Ansible plays
Go into the unixcraft-openshift repository directory
cd /tmp/unixcraft-openshift
Execute the following command
ansible-playbook -i inventories/openshift openshift.yml -b
Once the installation has finished, SSH into the master node and execute the following commands
htpasswd -c /etc/origin/htpasswd admin
You will be prompted to enter a password for the admin user
Then execute the following command
oadm policy add-cluster-role-to-user cluster-admin admin
Now the installation is complete and you are ready to logon to the Web Interface
Which should be found at:
https://osmaster01.lab.unixcraft.lcl:8443
Note: If the hostname of the master or the domain was changed, be sure to put in the correct hostname and domain.
You should be able now login as admin with the password provided to the htpasswd command
By default the installation does not mark the GlusterFS storage as default dynamic storage. To enable this we need to the following on the master node.
You need to create a default storage class which points to the existing GlusterFS storage. Lets get the details we will need to create a default storage class. Run the following commnd on the master node
oc get storageclass -o yaml
This will provide you with the existing storage class information. There are details in here we may need when creating our new default storage class.
Create a file with the following (name the file - default-storageclass.yml)
kind: StorageClass
apiVersion: storage.k8s.io/v1beta1
metadata:
name: generic
annotations:
storageclass.beta.kubernetes.io/is-default-class: "true"
provisioner: kubernetes.io/glusterfs
parameters:
resturl: http://heketi-storage-glusterfs.app.lab.unixcraft.lcl
restuser: admin
secretName: heketi-storage-admin-secret
secretNamespace: glusterfs
The details above should be correct, however to ensure we have the correct restuser, secretName, secretNamespace, compare the values with the ones given in the output of the command
oc get storageclass -o yaml
If all is good execute the following command:
oc create -f default-storageclass.yml
Now you have enabled the GlusterFS to be the default storage and be used for dynamic provisioning.
Note: Update from @jarrpa Simpler way to update / patch a storage class (example below)
oc patch storageclass generic -p '{"metadata": {"annotations":{"storageclass.kubernetes.io/is-default-class":"true"}}}'.