TLDR;
After nearly two years since its release, we have decided to decommission the EC2 (K3s) cluster feature.
Why this decision
The original goal of the EC2 (K3s) feature was to offer the same seamless Qovery experience on a “low-cost” infrastructure, particularly for development purposes.
However, following the rollout, we observed limited adoption. Most usage came from individual developers looking for a simple way to deploy side projects with a very high churn rate. Despite this, our broader target audience continued to prefer multi-node Kubernetes clusters, as the EC2/K3s setup lacked node autoscaling—an essential feature for many.
Additionally, maintaining compatibility for the EC2/K3s use case added complexity to our already intricate environment. Each code change required extra consideration for this feature, which consumed significant resources.
While it is disappointing to retire a feature we’ve invested considerable time in, we believe this is the right move. This decision allows us to refocus our efforts on more impactful areas, like enhancing support for Managed Clusters on Azure.
What will happen to the existing EC2(K3s) clusters
We’ve already removed the option to create new EC2 (K3s) clusters from our cluster creation flow.
Customers still using this cluster type have been informed and have successfully migrated their workloads to standard managed EKS clusters. We’ve recommended using Karpenter and spot instances to help optimize costs during this transition.
Have a great day
Alessandro