September 22, 2023
If you find yourself frequently rebuilding OpenShift clusters and potentially reusing cluster names, you may find it challanging to manage the credentials consistently and securely. Here is a solution using 1Password.
What’s in a Kubeconfig?
When you interact with a kubernetes cluster the
oc client will read and store configuration details in a file affectionately known as “kubeconfig” located at
$HOME/.kube/config. This file contains the CA certificates or
certificate-authority-data used to validate communication with a cluster. The file also keeps track of the usernames and namespaces you interact with. The combination of those items forms what is known as a “context”.
⭐ Pro Tip: See this post for details on extracting the certificate data from a kubeconfig to enable your operating system and web browser to trust your cluster.
During installation, OpenShift will create a system admin user and a kuebadmin user. The admin user
client-key-data are stored in the kubeconfig to enable passwordless authentication.
📓 Example kubeconfig file
- name: hub
- name: admin
- name: admin
The kubeadmin user authenticates with a password, which is not available in the kubeconfig. As a best practice this kubeadmin user is removed, but having the kubeadmin password readily available along side the kubeconfig can be very helpful.
It is important that a kubeconfig can be easily updated or replaced, because if the CA certificates are renewed or if you rebuild the cluster the
certificate-authority-data will be out of date, and the kubeconfig will no longer be able to verify the connection. For this reason, I maintain a distinct kubeconfig for each cluster I interact with. This is easily done by overidding the file location with the
KUBECONFIG environment variable.
So, how can you quickly and easily get your kubeconfig and kubeadmin credentials stored in 1Password?
Storing Kubernetes Credentials in 1Password
By uploading a kubeconfig and the kubeadmin password to a 1Password vault you can always have a clean copy of this data that can be easily refreshed and shared across your systems.
The following script uses the 1Password CLI to store or update the kubeconfig and kubeadmin credentials in a 1Password vault.
Start by pointing to the kubeconfig to be saved.
The script will upload that file and look for a
kubeadmin-password file in the same directory and upload that. The script will also login to the cluster to gather some other information such as the console URL, the cluster id, and creation date.
📓 There’s a bug here. When updating an existing entry. The kubeadmin password will be repeated. Do you know why?
Here is an example vault entry created by the script.
Reading Kubernetes Credentials from 1Password
Once stored in 1Password you can read the credentials and kubeconfig by sourcing an env file written like the following.
📓 Example env file to read kubernetes credentials from 1Password
I keep mine organized like this with a directory per cluster.
$ find ~/.kube/ocp -name .env
Select one of the above clusters by sourcing a corresponding env file like this.
$ source ~/.kube/ocp/hub/.env
$ echo $KUBECONFIG
$ oc whoami
Now you can “kube” to your ♥️ content!
If you redeploy the cluster, just remove the file from
~/.kube/ocp/<cluster>/kubeconfig and repeat above to update 1Password and write out a new kubeconfig.
There are other tools for managing kubernetes contexts but those seem to stem from putting too much information in one kubeconfig. I find this solution to be adequate for my needs.
As I mentioned there is a bug that duplicates the kubeadmin. Maybe you can fix it for me.
I suppose you could store the kubeconfig to a ramdisk instead of persisting it on the filesystem if you are extra paranoid. 🤷