A kubernetes operator for creating and managing a cache of container images directly on the cluster worker nodes, so application pods start almost instantly
Hello, thank you for the great repo. We plan to use this in our application however I have a few questions in mind.
I have a ML application and 2 node pools one is CPU-only pool and one is GPU-pool that scale down to zero, can we do cache on CPU-only node then move this cache to GPU node when a new node appear?
Where do kube-fledged store the image cache ? In the node disk, node memory or the persistent volume?
kube-fledged has an automatic cache refresh cycle. By default the cycle frequency is 15m. The frequency can be configured during deployment. Initially if the images are caches on the cpu-only nodes and later when a gpu node is added, the refresh cycle will automatically detect the newly added node and cache the images on those nodes.
kube-fledged stores the image cache in the node's disk
Hello, thank you for the great repo. We plan to use this in our application however I have a few questions in mind.
kube-fledged
store the image cache ? In the node disk, node memory or the persistent volume?