The sandbox.yaml reference in the containerd-env-setup.md document has issues

[whygyc]

Related issues: kubernetes-sigs/cri-tools#1273

The nydus-sandbox.yaml does not specify a uid which causes the crictl rmp to fail

metadata:
  attempt: 1
  name: nydus-sandbox
  namespace: default
log_directory: /tmp
linux:
  security_context:
    namespace_options:
      network: 2

As a test example, uid can be specified as an arbitrary string, like:

metadata:
  attempt: 1
  name: nydus-sandbox
  namespace: default
  uid: nydus-sandbox-test

This issue will be fixed in the new version, but it remains a problem for those using slightly earlier versions of crictl.

[imeoer]

Thanks for reporting, cloud you create a PR to fix it? thanks!

[whygyc]

https://github.com/dragonflyoss/nydus/blob/master/docs/containerd-env-setup.md

If you have encountered a similar issue, you can use ctr -n k8s.io c ls to view the same ID and use ctr -n k8s.io c rm <id> to delete it. After restarting the containerd service, it will no longer appear in crictl pods.

# crictl rmp 2aec11d869854
getting sandbox status of pod "2aec11d869854": metadata.Name, metadata.Namespace or metadata.Uid is not in metadata "&PodSandboxMetadata{Name:nydus-sandbox,Uid:,Namespace:default,Attempt:1,}"

[whygyc]

Thanks for reporting, cloud you create a PR to fix it? thanks!

ok, I will do it.