There are two strategies for upgrading the version of Kubernetes running on your Cloud 66 cluster:
- Without downtime: you swap out current servers individually for ones running newer versions until the entire cluster is updated - see the respective section below for instructions.
- With downtime: we update your entire cluster in parallel - this is much faster, but will bring down the entire cluster for a period of time - see the respective section below for instructions
Upgrading your cluster without downtime
It’s possible to upgrade both major and minor versions of Kubernetes running on your cluster without any downtime. This upgrade method revolves around scaling up new servers and swapping them out with the current servers (which are then decommissioned). As such it’s important to prepare thoroughly to avoid any possible downtime.
Don't skip versions
Never step up more than one major version level at a time - instead run this process multiple times.
Check your current version
First, confirm which version of Kubernetes your application is currently running. One easy way to do this is via the Cloud 66 Dashboard:
- Log into your Dashboard and click on your app
- Click on ⚙️ Settings in the left-hand nav
- Click on Show More under the Application & Servers panel
- Next to Application Type, click the More information link
- Scroll down to the Kubernetes Server panel and check the version displayed
You can also do this in the Manifest (see below) but confirming with your current live servers is the safest.
Update your Manifest
The version of Kubernetes used by new servers in your cluster is set via your manifest file. For example your app’s manifest might look as follows:
docker: configuration: kubernetes_version: 1.20.15 servers: ...
To set the next major version up you would change it to:
docker: configuration: kubernetes_version: 1.21.14 servers: ...
You made it!
Note that you can find Kubernetes version information here: https://kubernetes.io/releases/patch-releases/
Assess your current server capacity
Take careful note of the type, capacity and number of cluster servers - both nodes and masters. You will need to replicate this setup to fully upgrade your cluster.
Upgrade your cluster
- Ensure your Manifest is updated to the NEXT major version of Kubernetes (don’t jump more than one major version)
- Scale up sufficient new Kubernetes servers via your Dashboard (click the Servers tab and then the + Scale up Kubernetes button)
- Redeploy your app and test it carefully to ensure it’s working as expected
- Remove the older servers one by one, while monitoring your application for stability
You can then repeat the steps above for each major version required.
Upgrading with downtime
If you can afford to be tolerant of some downtime, you can upgrade multiple versions in one step which is much faster, and more fault tolerant. To do this:
- Add your desired version to your manifest
- Log into your Dashboard, and click on your app
- Click Deploy, then choose Deploy with Options
- In the drawer that slides out, click the Options tab and then select:
- Apply Docker/Kubernetes Upgrades (if Required)
- Use My Own Versions from My Manifest
- Clean Installation - Repair/Rebuild this Cluster
This will essentially recreate your entire cluster at the version you want.