During kube cluster bring up on a beta containerd codebase I was constantly seeing locked error when the kube infra pods were being created and run. Because of which cluster bring up was failing.
Sep 11 18:53:18 abhi-d-linux-1 cri-containerd[11225]: E0911 18:53:18.427543 11225 instrumented_service.go:38] RunPodSandbox for &PodSandboxMetadata{Name:kube-scheduler-abhi-d-linux-1,Uid:4fc47bcf2181edb3efb634af4abe6b09,Namespace:kube-system,Attempt:0,} failed, error: failed to get sandbox image "gcr.io/google_containers/pause:3.0": failed to pull image "gcr.io/google_containers/pause:3.0": failed to pull image "gcr.io/google_containers/pause:3.0": locking ref k8s.io/1/layer-sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4 failed: ref k8s.io/1/layer-sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4 locked: unavailable
The last time we tried this with @dmcgowan it seems to recover from the failure because kubelet recreates the failed pods. Opening this issue here to track it.
I will test this again and see if I hit this issue consistently on the latest codebase and post a more detailed log from kubelet and containerd. I will see if i can fix it even :)
During kube cluster bring up on a beta containerd codebase I was constantly seeing locked error when the kube infra pods were being created and run. Because of which cluster bring up was failing.
The last time we tried this with @dmcgowan it seems to recover from the failure because kubelet recreates the failed pods. Opening this issue here to track it.
I will test this again and see if I hit this issue consistently on the latest codebase and post a more detailed log from kubelet and containerd. I will see if i can fix it even :)