docs: add docs for running nydus snapshotter in Kubernetes #297
Conversation
| [plugins."io.containerd.grpc.v1.cri".containerd] | ||
| discard_unpacked_layers = false | ||
| disable_snapshot_annotations = false | ||
| snapshotter = "overlayfs" |
There was a problem hiding this comment.
Should we change the CRI snapshotter to nydus?
There was a problem hiding this comment.
Can we do that? First, we should pull the nydus-snapshotter image and start the container, then the nydus snapshotter can be used.
There was a problem hiding this comment.
I see. Nydus-snapshotter is deployed as DaemonSet and packed in an OCI container image.
Is it the same containerd pulling OCI image packing nydus-snapshotter and nydus image?
If yes, I don't understand how the containerd can choose nydus as snapshotter when pulling the nydus image ghcr.io/dragonflyoss/image-service/nginx:nydus-latest since containerd can only configure statically the snapshotter type here
There was a problem hiding this comment.
In the current helm chart, the snapshotter will be pulled and unpacked by overlayfs, and then change the global snapshotter to nydus and restart the containerd after the snapshotter started.
This new document will describe how to create a local Kubernetes cluster using kind and then deploy nydus snapshotter in the Kubernetes cluster. Signed-off-by: bin liu <liubin0329@gmail.com>
liubin
force-pushed
the
docs/run-in-k8s
branch
from
97932e9
to
2ced0f8
Compare
This new document will describe how to create a local Kubernetes cluster using kind and then deploy nydus snapshotter in the Kubernetes cluster.
Signed-off-by: bin liu liubin0329@gmail.com