Docker Desktop Kubeadm Storage Errors

Docker Desktop
,
1

After updating to Docker Desktop 4.38.0 for Windows 10, enabling Kubernetes by selecting Kubeadm, the cluster is successfully deployed (version 1.31.4). Pods also appear to successfully deploy to cluster, and can be listed with kubectl get pods. However, when attempting to exec into pods with kubectl, the following error is presented:
OCI runtime exec failed: write /tmp/runc-process[xxxx]: no space left on device: unknown.

PVs also create successfully, but pod logs indicate that issues are present in writing to these PVs, also displaying an “out of space” error.

I am using the default WSL docker-desktop backend. Previously, when I deployed containers requiring persistent storage, I used the “hostpath” storage type, creating mountpoints at /data inside of the wsl distro. I also no longer see these mount points fill with data, despite the PVs successfully creating and the PVCs showing successful binds.

I have attempted resetting the k8s cluster and re-installing Docker Desktop, neither of these attempts resolved the issue.

Has anyone else experienced this? Any suggestions on resolution are appreciated.

2

If you still have the issue:

Is it possible you have no space left under the VM image? Or that Windows made it read-only and somehow it detected as no space left? Of course, pods could not start if the pod metadata files are stored on the same filesystem.

You can try to enable system containers on the Kubernetes tab in settings and use docker exec to look around in a container. You could do it with using docker run, but using a Kubernetes pod when you have problem with kubectl commands is a better test.

I would run df -h in a container.

update:

Forget system containers. It is not needed for showing other pods.

3

This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.