A tool to provision and install OpenShift clusters.
Usage:
Usage: main.py [OPTIONS] COMMAND [ARGS]...
Options:
--help Show this message and exit.
Commands:
create Provision + Install + Setup
destroy Destroy infrastructure
install Install OpenShift on infrastructure
provision Provision infrastructure
setup Run post-installation setup
See examples
directory.
Create an empty directory. Add public and private SSH key that will be used to connect to the provisioned machines. If you do not want to reuse existing key, simply generate new one
$ ssh-keygen -f openshift-key -N ""
and you will get two files mykey
and mykey.pub
.
If you want to reuse existing keys, make sure they're unencrypted or have empty password.
Create yaml definition for your cluster as defined above and name the file as the cluster should be named, e.g.
cluster01.yml
.
Check the provider documentation for provider-specific files and configuration to add to the definition.
Start the deployment process
$ docker run -ti -v (path to your directory):/root/data docker.io/osevg/openshifter create cluster01
NOTE: If running docker
directly from the same folder where descriptors and SSH keys are located, you can pass in pwd
as path to your directory, e.g.
$ docker run -ti -v `pwd`:/root/data ...
Following are instructions required to get OpenShift working on top Google Cloud Platform (GCP).
First install the [Google Cloud SDK](register for an account in Google Cloud ()) in your local environment. Then, get a GCP account if you don't already own one.
Next, you need to create a project for running OpenShift on it.
Creating the project can be done either using the GCP user interface or via the gce/create-project-gce.sh
script.
The instructions for using the user interface are not available yet, so the best would be to use the script.
To run the script, you just need to provide a name and the email registered with your GCP account:
./gce/create-project-gce.sh openshift-on-google me@here.com
Calling this script will generate a project on GCP called openshift-on-google-me-here-com
.
The script also generates openshift-on-google-me-here-com.json
file which is needed for OpenShifter to talk to GCP.
This JSON file name, along with the generate project name, need to be added to the cluster yaml file:
gce:
account: os-v1-me-here-com.json
region: us-west1
zone: us-west1-a
project: os-v1-me-here-com
Next, call create on OpenShifter as shown in the general instructions above.
When not using OpenShift instance on Google Cloud, make sure you shutdown cluster master and other nodes.
To do that go to Compute Engine / VM instances
and stop all VM instances.
In this section you will find instruction on things to try if things are not working:
First, try to remove all OpenShifter images locally and get a fresh one, e.g.
docker ps -a | awk '{ print $1,$2 }' | grep openshifter | awk '{print $1 }' | xargs -I {} docker rm {}
docker rmi osevg/openshifter
If that does not solve your problem and you have a partially installed cluster, try destroying it:
docker run -e -ti -v (path to your directory):/root/data docker.io/osevg/openshifter destroy cluster01
Note that at times, destroying a cluster might timeout. If that happens, simply try again until it completes.
Finally, you can find out more information about what OpenShifter does by passing in DEBUG=true
as environment variable:
docker run -e "DEBUG=true" -ti -v (path to your directory):/root/data docker.io/osevg/openshifter create cluster01
This section explains common errors found and how to resolve them:
When trying to create the OpenShift cluster, you might see an error like this:
paramiko.ssh_exception.SSHException: not a valid OPENSSH private key file
This error is complaining about the SSH keys generated above.
There's no need to generate OPENSSH
keys here, the RSA
keys generated above should work.
Normally, this error appears when trying to connect to a partially installed OpenShift cluster which maybe used a different key to the one you're passing in now.
So, to overcome this problem, simply destroy the OpenShift cluster and recreate it:
docker run -e -ti -v (path to your directory):/root/data docker.io/osevg/openshifter destroy cluster01
docker run -e -ti -v (path to your directory):/root/data docker.io/osevg/openshifter create cluster01
The generated SSH keys are password protected.
Either unencrypt them or regenerate them with an empty password.
You can pass in -N ""
to ssh-keygen
to automate this.
If you get this error, the yaml file contains a non-ASCII character. To find the offending line(s) call:
perl -lne 'print if /[^[:ascii:]]/' cluster01.yml
This error can appear when your computer has gone to sleep and the Docker VM's clock got out of sync. Restarting the Docker daemon should fix it.