You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When executing the lifecycle inside a kubernetes pod we specify the pod's securityContext.fsgroup to mount volumes as the CNB_USER. This will be result in the volume mounts for /layers and /cache owned by root with the group specified fsGroup.
The EnsureOwner lifecycle check will still attempt to chown the volume mounts on these volumes because it checks both the uid and gid. This results in an error on analyze:
failed to chown volumes: chown /layers: operation not permitted
The text was updated successfully, but these errors were encountered:
I think we can assume that, if the gid, is correctly, lifecycle doesn't need to fix the permissions. This shouldn't break the pack case when we need to chown the volumes, but should fix the permissions error you are seeing in k8s.
After testing #272 on minikube, it seems that not all storage classes will accurately respect the securityContext.fsgroup. Is it possible to just check that those directories are writeable?
When executing the lifecycle inside a kubernetes pod we specify the pod's
securityContext.fsgroup
to mount volumes as the CNB_USER. This will be result in the volume mounts for/layers
and/cache
owned by root with the group specifiedfsGroup
.The EnsureOwner lifecycle check will still attempt to chown the volume mounts on these volumes because it checks both the uid and gid. This results in an error on analyze:
The text was updated successfully, but these errors were encountered: