Add-ons

Install, configure, and remove one-click add-ons on a cluster, and understand their dependencies.

Add-ons are curated components you can install on a cluster with one click from the Resource tab. Versions are pinned and tested by the platform; you choose what to install and how it is configured.

The catalog

Add-on	Category	Purpose
Metrics Server	Infrastructure	Resource metrics for `kubectl top` and autoscaling. Always installed; cannot be removed.
cert-manager	Infrastructure	Issues and renews TLS certificates inside the cluster.
Traefik Ingress	Infrastructure	Ingress controller; required for HTTP routes.
HAProxy Ingress	Infrastructure	Alternative ingress controller.
Longhorn	Infrastructure	Distributed, replicated block storage across worker data disks.
OpenEBS LVM LocalPV	Infrastructure	LVM-backed local persistent volumes on worker data disks.
Local Path Provisioner	Infrastructure	Simple node-local persistent volumes.
KubeVirt	Infrastructure	Run virtual machines as cluster workloads.
Prometheus	Observability	Prometheus Operator plus node and cluster-state exporters (the Prometheus server itself is disabled by default).
Loki	Observability	Log aggregation.
Grafana Alloy	Observability	Telemetry collector that ships cluster logs to Loki.
Valkey	Database	In-memory key-value store.
CloudNativePG	Workload Operator	PostgreSQL operator behind Akasha PSQL.
Flux CD	Infrastructure	GitOps controllers; required for the GitOps tab.
Zot Registry	Infrastructure	Per-cluster OCI container registry.
GitHub Actions Runner Controller	CI/CD	Operator for self-hosted GitHub Actions runners.
GitHub Actions Runner	CI/CD	A runner scale set; prompts for your GitHub repository or organization URL and a personal access token.

Dependencies

Some add-ons depend on others. The catalog card shows a Requires: note and blocks installation until the dependency is met:

Stateful add-ons (KubeVirt, Loki, Valkey, CloudNativePG, Zot) require a storage add-on first — Longhorn, OpenEBS LVM, or Local Path Provisioner. Note: Loki's default configuration pins its storage to the longhorn storage class; if you use a different storage add-on, change singleBinary.persistence.storageClassName in the Loki values.
Grafana Alloy requires Loki.
GitHub Actions Runner requires the GitHub Actions Runner Controller.

Install an add-on

Open the cluster and select the Resource tab.
Click Install Add-on to open the catalog.
Pick a card and click Install. Add-ons with required inputs (such as the GitHub Actions Runner) prompt for them first.

Installs run as queued operations. The add-ons table shows the state: Pending while the install is in flight, Deployed/Ready on success, and Failed with a Retry install button on error.

Feature pages that depend on add-ons state their requirement: VMs requires KubeVirt, and Databases requires CloudNativePG.

Configure an add-on

Click an installed add-on to open its configuration panel:

Values — a YAML editor of the add-on's live Helm values. Saving applies an upgrade as a queued operation.
History — past revisions, with a Rollback action per revision.
Manifest Diff — compare the rendered manifests between two revisions.
Settings — chart, release, and namespace details, and a Remove action.

Add-ons managed through GitOps are read-only here — the panel shows "Managed by Flux", and changes (including removal) must go through the Git source.

Remove an add-on

Use Remove in the add-on's Settings panel. The release is uninstalled from the cluster. Metrics Server cannot be removed.

What the table reflects

The add-ons view tracks components installed through Akasha, alongside Helm releases it detects on the cluster. Software you install by other means may not be recognized as an installed add-on, and dependency checks only count add-ons Akasha knows about.

Create a cluster — bundles pre-select common add-on sets
Ingress and external IPs
GitOps with Flux
Virtual machines on Kubernetes