There are a couple different ways this could be useful to us. We currently use minikube, kubernaut, and long-lived clusters for a variety of scenarios related to development and CI. Initial investigation seems to suggest k3s with a relatively small amount of work could provide improvements over the alternatives in some cases.
Currently it appears to function as a drop-in replacement for minikube and kubernaut in dev scenarios, modulo lack of image caching. With a bit of work on the registry/image caching front it could provide a potentially improved dev experience, e.g. nearly fully local dev.
On the CI front it could potentially provide a nice consistent way to do testing without credentials and without exposing our own cloud resources the way kubernaut does.
The primary outstanding issue seems to be a reasonable way to get docker images into k3s that doesn't require a full remote roundtrip with credentials, and doesn't throw away caching whenever k3s is restarted.
There are a couple different ways this could be useful to us. We currently use minikube, kubernaut, and long-lived clusters for a variety of scenarios related to development and CI. Initial investigation seems to suggest k3s with a relatively small amount of work could provide improvements over the alternatives in some cases.
Currently it appears to function as a drop-in replacement for minikube and kubernaut in dev scenarios, modulo lack of image caching. With a bit of work on the registry/image caching front it could provide a potentially improved dev experience, e.g. nearly fully local dev.
On the CI front it could potentially provide a nice consistent way to do testing without credentials and without exposing our own cloud resources the way kubernaut does.
The primary outstanding issue seems to be a reasonable way to get docker images into k3s that doesn't require a full remote roundtrip with credentials, and doesn't throw away caching whenever k3s is restarted.