I'm trying to start couchdb in a container in Kubernetes on GKE but the start fails:
Normal Pulled 10m kubelet Successfully pulled image "couchdb:3.1.1" in 574.834697ms (574.8564ms including waiting)
Warning Failed 10m (x2 over 11m) kubelet Error: failed to create containerd task: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error setting cgroup config for procHooks process: openat2 /sys/fs/cgroup/kubepods.slice/kubepods-pod0ee23bc6_76f5_4927_89c3_c8d4a71ff30f.slice/cri-containerd-couchdb.scope/cpu.weight: no such file or directory: unknown
I'm using an auto pilot cluster and this is the only image which fails. What can be the reason?
The error is very misleading. In GKE (auto pilot) only the resource and not the the limit is considered and summed up to the next possible value if too low. When having a deployment with 2 defined containers in my case this results in giving the container 1 240 Mi and the second 10Mi since the ressource values defined by the helm chart I'm using are too low (just specifying 10Mi for each container). Increasing the resource values solved this problem, so it is not related at all to the reported cgroup error.