-
Notifications
You must be signed in to change notification settings - Fork 511
Description
Is your feature request related to a problem? Please describe.
Currently, the MinIO Tenant Helm chart hardcodes the accessModes for PersistentVolumeClaims directly in the chart templates. This makes it impossible to adapt the access mode to specific storage backends or Kubernetes distributions.
On OpenShift, this causes a significant issue: when MinIO pods start, the PVCs trigger a full SELinux relabeling process. In our cluster, this results in pod startup times exceeding 30 minutes, which is highly disruptive for operations and scaling. (e.g https://access.redhat.com/solutions/7079407)
Describe the solution you'd like
I would like to be able to configure the accessModes of the PVCs to RWOP through the Helm values (for example via values.yaml), instead of having them hardcoded in the chart templates.
This would allow users to choose an access mode that better fits their storage class and platform constraints (e.g. OpenShift), and avoid unnecessary SELinux relabeling delays.
Describe alternatives you've considered
Forking and maintaining a custom version of the MinIO Tenant Helm chart with modified PVC templates
Manually patching the generated PVCs after deployment
Both approaches are error-prone, hard to maintain, and break the Helm upgrade workflow.
Additional context
This issue specifically impacts OpenShift clusters where SELinux relabeling can be very time-consuming depending on the access mode and storage backend.
Making accessModes configurable would greatly improve flexibility and startup performance for MinIO tenants on such platforms.