Open davidpanic opened 1 year ago
this process is documented here: https://www.talos.dev/v1.4/kubernetes-guides/configuration/ceph-with-rook/#talos-linux-rook-metadata-removal
Adding delete api conflicts with talos philosophies.
We might add some structured way to perform partial wipe of /var
.
If this cluster is HA, you can wipe whole /var
with talosctl reset --system-labels-to-wipe=EPHEMERAL --reboot
Hi,
I've created a linstor snapshot as a test, then deleted it and I'm not stuck with the underlying file in the middle of my volumes. An easy way to delete a file in /var would be useful, I guess my only option now is to figure out how to execute a privileged pod with access to that path to delete the file from there
@Ulrar This might come in useful, I'ts what I used to delete the rook directory:
apiVersion: batch/v1
kind: Job
metadata:
name: cleanup-rook-dir
spec:
ttlSecondsAfterFinished: 0
template:
spec:
nodeName: talos-test-1
containers:
- name: cleanup
image: alpine:latest
command: ["rm", "-rf", "/mnt/rook"]
volumeMounts:
- name: var-lib
mountPath: /mnt
restartPolicy: Never
volumes:
- name: var-lib
hostPath:
path: /var/lib
This issue is stale because it has been open 180 days with no activity. Remove stale label or comment or this will be closed in 7 days.
Yeah, the workaround works, but it's still pretty annoying to do when there could just be an API to do it. Commenting to remove stale label.
Currently there is no way that I know of to delete or write a specific file or directory on the nodes. The only existing file manipulation APIs are
read
andlist
but there is nowrite
ordelete
.I would like to have this functionality to be able to remove
/var/lib/rook
as it is required for correct cleanup of a Rook cluster.To work around this I have just run a pod with an alpine container and a local volume, then deleted the directory from within the pod.