-
Notifications
You must be signed in to change notification settings - Fork 6
Open
Description
Thanks for sharing your work within the community !
We're running kubernetes k3s disto on top of nodes managed by cloudfoundry bosh infrastructure, see https://bosh.io/ and https://github.com/orange-cloudfoundry/k3s-wrapper-boshrelease for more background.
In our environment, the kubernetes nodes are reached by ssh using a guest user which has sudo/su permissions, but whose account is not configured for user interaction (e.g with crictl in path)
server/11aa3763-7a95-47ad-8089-7399a97a2d0d:~/test# ./kubectl-nsenter --use-node-name=false --user guest -n kube-system coredns-66f786f4cb-ghcvk /coredns
time="2025-07-10T10:07:06Z" level=fatal msg="cannot access containerd socket \"/run/containerd/containerd.sock\": no such file or directory"
nsenter: failed to parse pid: '/coredns'
./kubectl-nsenter: failed to execute ssh command: failed to run child process "ssh": exit status 1server/11aa3763-7a95-47ad-8089-7399a97a2d0d:~/
| exec crictl inspect --output go-template --template={{.info.pid}} %[1]s; |
In our environment, the following sequence succeeds
ssh guest@192.168.116.224
sudo -i
crictl inspect --output go-template --template={{.info.pid}} 028df9c85f9db;
218035
Would it make sense to support such environments, by adding the possibly to sudo -i prior to run the discovery commands ?
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
No labels