This message was deleted.
# general
a
This message was deleted.
c
Being aware of the namespace stuff is a good start, but are you sure that the image you’re building has the same name as the image you’re referencing in the pod?
Are you also sure that you’re talking to the same containerd instance as the k3s kubelet?
e
I believe the name in the yaml file for the deployment is the same name as the image I am building.
Hmm. As part of this setup I had to enable containerd as "systemctl is-enabled containerd.service" was returning false and nerdctl build was complaing. The K3s install however was already working though. Perhaps there is more than once instance of containerd?
c
oh yeah, you’re running two different instances of containerd then
you should shut down and uninstall the system containerd, and point nerdctl at the k3s containerd socket
e
ok, thanks. I did that and I'm back to "cannot access containerd socket "/run/containerd/containerd.sock"" so then I tried adding --address /run/k3s/containerd/containerd.sock to nerdctl build and well, it gets further but that still fails with errors about unix:///run/containerd/containerd.sock
There must be something else I am missing here.
c
at what point does it indicate that it’s still looking at the wrong socket address?
you might try using
--address unix:///run/k3s/containerd/containerd.sock
instead? sometimes its fussy about that
you could also just symlink /run/containerd/containerd.sock to /run/k3s/containerd/containerd.sock 🤷
e
Well, that seems to work. Seems like I should not have to do that once I specify the socket address though. But thanks! This got me unstuck. I'll report to nerdctl.