raff kubernetes

_{Updated August 24, 2026} The raff kubernetes command group (alias: raff k8s) covers every managed Kubernetes operation in the public API. Each subcommand supports --output json for scripting.

Subcommand index

Subcommand	What it does
`list`	List all clusters in the project
`get`	Show details for one cluster
`create`	Create a cluster
`delete`	Delete a cluster and all its infrastructure
`rename`	Rename a cluster
`kubeconfig`	Print or save the cluster’s kubeconfig
`nodes`	List the cluster’s nodes with live status
`pool list/add/scale/delete`	Manage node pools
`upgrade-ha`	Promote to a 3-master HA control plane
`upgrade`	Upgrade the Kubernetes version (rolling, in place)
`maintenance`	Set the upgrade mode and maintenance window
`versions`	List available Kubernetes versions
`plans`	List worker node plans with pricing
`volumes`	List the cluster’s PVC-provisioned volumes

list

raff kubernetes list [--output table|json]

get

raff kubernetes get <cluster-id> [--output table|json]

Shows status, version, endpoint, HA, node counts, and pricing for one cluster.

create

raff kubernetes create --name <name> --plan <plan-id> [--nodes 2] [--version <version-id>] [--ha] [--wait] [--idempotency-key <key>]

Flag	Description
`--name`	Cluster name (1–63 lowercase letters, digits or hyphens). Required.
`--plan`	Worker node plan ID — see `plans`. Required.
`--nodes`	Worker node count for the default pool (2–20; added pools can be 1–20 with their own plan). Default 2.
`--version`	Kubernetes version ID — see `versions`. Default: platform default.
`--ha`	HA control plane (3 masters, flat monthly fee).
`--wait`	Block until the cluster is running (10–20 minutes).
`--idempotency-key`	A retried create with the same key returns the same cluster.

Creates follow the account’s billing mode: pay-as-you-go bills hourly; on a subscription account the CLI create charges the saved payment method for the monthly subscription automatically.

raff kubernetes create --name prod-cluster --plan 1 --nodes 3 --ha --wait

delete

raff kubernetes delete <cluster-id> [--force] [--wait]

Deletes the cluster and all its infrastructure — nodes, gateway, VPC, and floating IP. --force skips the confirmation prompt; --wait blocks until deletion completes.

rename

raff kubernetes rename <cluster-id> <new-name>

The API endpoint and DNS stay the same.

kubeconfig

raff kubernetes kubeconfig <cluster-id> [--save <path>] [--ttl <duration>] [--static]
raff kubernetes kubeconfig rotate <cluster-id>

Writes a self-renewing kubeconfig: the file carries no credential, and kubectl calls the Raff CLI for a fresh short-lived token before each request. It never expires, nothing long-lived is stored, and the file is useless to anyone without your API key. --ttl sets each token’s lifetime (default 1h). --static embeds a token in the file instead, for environments that cannot run the CLI. It expires (7 days by default, --ttl to change, 30 days maximum) and must be re-issued afterwards. kubeconfig rotate invalidates every kubeconfig and token issued for the cluster, immediately.

raff kubernetes kubeconfig zk7x9m --save ~/.kube/raff-zk7x9m.yaml
export KUBECONFIG=~/.kube/raff-zk7x9m.yaml
kubectl get nodes

nodes

raff kubernetes nodes <cluster-id>

Lists every node (gateway, control plane, workers, storage) with role, status, and IP.

Node pools

raff kubernetes pool list <cluster-id>
raff kubernetes pool add <cluster-id> --name <name> --plan <plan-id> [--nodes 2]
raff kubernetes pool scale <cluster-id> <pool-id> <node-count>
raff kubernetes pool delete <cluster-id> <pool-id> [--force]

Scale-down and pool deletion drain nodes first — workloads are evicted respecting PodDisruptionBudgets before a node is removed. Each pool has its own worker plan — pass any plan ID from plans to mix sizes and tiers (General Purpose / CPU-Optimized) in one cluster. Added pools can have 1–20 nodes; the cluster keeps at least 2 workers overall.

upgrade-ha

raff kubernetes upgrade-ha <cluster-id> [--force]

Promotes a single-master cluster to a 3-master HA control plane with a redundant gateway pair. Cannot be reversed.

upgrade

# List available targets
raff kubernetes upgrade <cluster-id>

# Start an upgrade
raff kubernetes upgrade <cluster-id> --version <version-id> [--confirm-single-master]

Rolling in-place upgrade: control-plane nodes one at a time, then workers drained one at a time. On a non-HA cluster the Kubernetes API is briefly interrupted — pass --confirm-single-master to accept that. Minor versions upgrade sequentially; upgrades cannot be rolled back. Progress appears in the cluster’s Activity feed; raff kubernetes get shows the version once complete.

maintenance

raff kubernetes maintenance <cluster-id> [--mode manual|auto_patch|auto_minor] [--day 0-6] [--start 0-23]

Flag	Description
`--mode`	`manual` — you start upgrades. `auto_patch` — patches apply automatically in the window. `auto_minor` — minors too, after a stability period.
`--day`	Maintenance day: 0 (Sunday) to 6 (Saturday).
`--start`	Window start hour (UTC); the window is 4 hours.

Only the flags you pass change; the rest keep their values.

versions

raff kubernetes versions

Lists the Kubernetes versions offered, with the version ID used by create --version. Creating on a version newer than the base image works — the cluster upgrades itself to it during creation.

plans

raff kubernetes plans

Lists worker node plans with class (General Purpose shared / CPU-Optimized cpu_optimized), vCPU, memory, and monthly price. Every worker has a 25 GiB root disk.

volumes

raff kubernetes volumes <cluster-id>

Lists the cluster’s PVC-provisioned persistent volumes (the raff-block StorageClass): PV name, size, status, the worker it is attached to, and the hourly price.

Getting Started

Guides

REST API

CLI

Terraform

raff kubernetes

Subcommand index

list

get

create

delete

rename

kubeconfig

nodes

Node pools

upgrade-ha

upgrade

maintenance

versions

plans

volumes

​Subcommand index

​list

​get

​create

​delete

​rename

​kubeconfig

​nodes

​Node pools

​upgrade-ha

​upgrade

​maintenance

​versions

​plans

​volumes

Subcommand index

list

get

create

delete

rename

kubeconfig

nodes

Node pools

upgrade-ha

upgrade

maintenance

versions

plans

volumes