Cluster name, deployment name, and the unique suffix

qumulo_filesystem_aws, qumulo_filesystem_azure, and qumulo_filesystem_gcp each name a cluster twice, once for people and once for the cloud. This page is about what each name controls, why the cloud name has a per-cloud limit, and the suffix the provider adds to it.

Two names, two audiences

cluster_name is the qfsd cluster name shown in the Qumulo web UI. It keeps its case and is limited to 2 to 15 characters.

deployment_name seeds the names of the cloud resources, such as instances, disks, buckets, and roles. Cloud resource names must be lowercase, so the provider lowercases it, and changing only its case renames nothing.

resource "qumulo_filesystem_aws" "example" {
  cluster_name    = "Prod-Data"     # shown in the Qumulo web UI
  deployment_name = "acmestorage01" # seeds cloud resource names
}

The name attribute sets both at once, using the value as written for the cluster name and in lowercase for the deployment name. It carries the cluster name's 15-character limit, and the provider rejects name set alongside either of the other two.

Each cloud caps the deployment name

Every cloud limits the length of some resource names, and the deployment name has to fit inside them with its suffix. It takes letters, digits, and interior hyphens.

Cloud	deployment_name limit
AWS	36 characters
Azure	15 characters
GCP	16 characters

Azure's cap comes from its 24-character storage account name limit, and GCP's from its 30-character service account name limit.

A deployment name longer than 15 characters needs the cluster_name and deployment_name pair, since name cannot exceed 15.

The suffix makes the name unique

On the first deploy, the provider appends a random six-character suffix of lowercase letters and digits to the deployment name. The result is deployment_unique_name, such as acmestorage01-h12g9v.

The unique name, not deployment_name, names and tags every cloud resource and every cluster node. It never changes for the life of the deployment, so nodes added by a later scale share it with the originals.

Automation that needs the real resource names reads the computed deployment_unique_name attribute. It is also the last segment of the cluster's import ID, as Import an existing cluster shows.

Changing deployment_name to a different name, rather than a different case, creates a new cluster with a new unique name.