Now, it's not possible to provision new worker nodegroups in the existing cluster, because slurm-cluster-storage Helm chart has nodeAffinity with the specific list of all nodegroups. The problem is that field is immutable in K8s for PVs. So in order to support new nodegroup provisioning, we have to refer to some label that is common for all worker nodegroups, such as slurm.nebius.ai/nodeset=worker in addition to existing ones slurm.nebius.ai/nodeset=worker-0, slurm.nebius.ai/nodeset=worker-1, etc.