Storage

Storage capacity can grow but cannot shrink, even if your data would fit in less. This applies to cluster storage and to resizing disks and Kubernetes volumes.

On a Slurm cluster, configured shared storage backs /home and /shared on the login and compute nodes. Ceph and NFS can provide that shared filesystem. The inventory also reports WEKA when configured. See Connect over SSH.

Increase storage capacity

Each storage purchase adds to the cluster’s capacity rather than setting a new total. The purchase object below and storage.purchase in cluster options give the Business API route for buying more storage for a cluster. Purchases made there go through the usual quote and purchase checks.

Backends and capacity

A cluster can have Ceph, WEKA, or both. NFS is used on its own.
unknown means the cluster has not reported whether that backend is configured. Read again later.
purchasedRawBytes is the total storage purchased, not the space used or available. configuredQuotaBytes is the Ceph quota the cluster has applied. The two can differ for a short time after a purchase.
Only Ceph entries have quota fields; their absence on WEKA and NFS entries does not mean zero capacity. For NFS readiness, read storage.ready from cluster options.
Slurm’s shared volume requests 100 GiB when no size is set. That request is separate from the purchased capacity and the Ceph quota.

Read organization storage

GET /organizations/{orgId}/storage

API: Platform. Permission: clusters.read in the organization. If you only have a grant for one cluster, read that cluster’s options instead. Returns one entry per cluster the organization owns, including clusters with no storage. The list is not paginated.

Parameters

Name	In	Type	Required	Description
`orgId`	path	string	yes	Organization ID.

Response

Field	Type	Description
`items[].clusterId`	string	Cluster ID.
`items[].clusterName`	string	Cluster name.
`items[].lifecycleState`	string	The cluster’s lifecycle state. See the cluster object.
`items[].state`	string	`available`, `unavailable`, or `unknown`, as for cluster options.
`items[].reason`	string, optional	Why `state` is not `available`.
`items[].observedAt`	timestamp, optional	When the cluster was last observed.
`items[].backends[].type`	string	`ceph`, `weka`, or `nfs`. Ceph and WEKA can coexist; NFS excludes both.
`items[].backends[].configured`	string	`configured`, `not_configured`, or `unknown`.
`items[].backends[].configuredQuotaBytes`	integer, optional	The Ceph quota the cluster reports, in bytes. Absent for WEKA and NFS.
`items[].backends[].purchasedRawBytes`	integer, optional	Total bytes of active storage purchases, on the Ceph entry. Not the space used or available.
`items[].backends[].ready`	condition, optional	Ceph readiness. See conditions.
`items[].backends[].purchase`	object, optional	How to buy more storage for this cluster, as in cluster options. On the Ceph entry of an active or provisioning cluster.

Example

andromeda call platform GET "/organizations/$ORG_ID/storage" |
  jq -r '.items[] | .clusterName as $c | .backends[]
    | [$c, .type, .configured, (.configuredQuotaBytes // "not reported")] | @tsv'

training-h200	ceph	configured	1000000000000
training-h200	weka	not_configured	not reported
training-h200	nfs	not_configured	not reported

Errors

Status	Code	When
403	`forbidden`	You lack `clusters.read` in the organization.

Overview

Business API

Platform API

Increase storage capacity

Backends and capacity

Read organization storage

Parameters

Response

Example

Errors

​Increase storage capacity

​Backends and capacity

​Read organization storage

​Parameters

​Response

​Example

​Errors

Increase storage capacity

Backends and capacity

Read organization storage

Parameters

Response

Example

Errors