In ROSA-HCP clusters, the ETCD database size configuration appears to be misaligned with expectations. Although the persistent volumes (PVs) provisioned are correctly set to 32Gi in size, the ETCD instance itself is not utilizing the full allocated capacity. The QUOTA_BACKEND_BYTES environment variable, which sets the maximum size of the ETCD backend database is set to 8Gi, regardless of the actual PV size.

Also the ETCD metric etcd_server_quota_backend_bytes, which consistently reports a value of 8Gi. Overriding on self-managed hypershift cluster thru '--etcd-storage-size' arg is not updating the actual DB limit.

    

All recent versions
    

Always
    

    1. Create a ROSA-HCP
    2. Check its PV and PVC capacity on Management Cluster
    3. Check etcd_server_quota_backend_bytes or etcd container environment variable
    

ETCD still runs with limited 8Gi space even for larger cluster.
    

ETCD DB to configured it full capacity