During the #KUBE100 beta programme, when your monthly invoice is issued, a discount of $70 or the invoice balance will be applied (whichever is lower). If your usage for the month through clusters, instances or other Civo resources exceeds $70 in a given month, the overage will be charged to the card on file.
You can set up a billing alert to notify you if you are approaching a particular value. We can also adjust your account quota (of deployable instances, volumes, etc) on request to prevent the likelihood of you exceeding your budget.
Yes. Our managed Kubernetes service runs on k3s, which due to its small footprint allows the master node to also perform work alongside the other nodes.
We have written a comprehensive CLI tool to manage virtual machine instances, Kubernetes clusters and anything else to do with your Civo account. You can find out how to set it up here.
If you do not want to set up Ruby to run the CLI
gem, you can run it as a Docker Alias and map your API key to the alias with these instructions.
To provide consistency and predictability of service for Civo managed Kubernetes, SSH connections to the cluster nodes are not possible. However, there is nothing stopping you from creating a Kubernetes cluster composed of Civo instances that you manage yourself, and allowing SSH connections to these nodes.
Another reason there is no SSH access is that we want to keep our options open about moving the architecture of the managed Kubernetes service to a virtualised platform such as
k3v or similar in the future, and thus want to make sure users will experience a consistent service.
Related to the question above, because the nodes themselves are not accessible via SSH or managed outside the cluster, they will not appear as separate instances. If you need to restart a particular node, you can click on the "recycle" button on the cluster interface page.
k3s starts with
Traefik as an Ingress Controller. This is to allow connections to your cluster. You can prevent Traefik from being installed when you start a cluster by deselecting it in the web UI (under 'Architecture') or by adding
--remove-applications=traefik to your Civo CLI cluster creation command.
Remember that starting up a cluster with no Ingress Controller means you will manually have to configure one to allow access to applications.
Only the master node has a public IP address. It receives traffic to your cluster, and routes it to your worker nodes in the form of a load balancer, as well as processing tasks itself.
We aim for the underlying operating system to be as lightweight as possible to allow for maximum resources to be allocated to Kubernetes and your applications. To this end, we may change the underlying OS, but as this is a managed service this will be the same for all users.
We are planning to implement pooling of nodes of different sizes (e.g. 2x medium, 2x large) in the same cluster, and will annouce when this feature is available.
We would love to see you document your experience, whether it is on your own site, on Medium/DEV.to or as a Civo learn guide. If you write up a piece, submit a pull request at the Civo Community Content to make sure we know about it.
Our CTO Andy wrote a great post about precisely this topic, and why we decided to go with Rancher's k3s as the Kubernetes distribution of choice.
k3s is fully compatible with upstream Kubernetes (K8S).
Take a look at this blog post that explores use cases and the advantages
k3s gives you. We are also keen to hear from you about your use cases! See "Can I contribute content" above.