Lab · HA ready

Pertisk KOS

Immutable Kubernetes OS, API-only.

Pertisk KOS boots a locked-down Linux image whose only job is to run Kubernetes nodes. PID 1 is pertiskd in Rust: disk, net, containerd, kubelet, and signed A/B updates — without a shell on the machine. The management UI provisions clusters on Proxmox, standalone ESXi, and Nutanix AHV (Prism Element). Cluster add-ons cover NFS, cert-manager, Cilium LoadBalancer, Pertisk Ingress, and kos-scaler worker VM autoscaling.

Documentation GitHub repository Live demo All projects

Pertisk KOS management dashboard with cluster resource cards

Highlights

API-only management with mTLS (pertiskctl)
Immutable root with signed A/B image updates
Mgmt UI: Proxmox, ESXi, and Nutanix AHV — dashboard, cluster ops, add-ons
Same API from the UI, Terraform, or pertiskctl

Stack

Rust
containerd
kubelet
gRPC
React mgmt UI
UKI / Secure Boot

Management UI

Hash routes on the live demo. Roles: viewer (read), operator (mutate clusters), admin (users + delete providers).

Login

/#/login

Local username/password, optional Auth0 SSO, remember password, forgot password.

Sign in
Continue with Auth0 SSO
Forgot password?

Dashboard

/#/

Cluster stats and cards (CPU / memory / disk) stacked above provider cards. Recent clusters table.

Create cluster
All clusters
All providers
Refresh
Open a cluster or provider card

Clusters

/#/clusters

All clusters: status, arch, provider, topology, resource bars, network, CNI.

Create cluster
Open a row for cluster detail

Create cluster

/#/clusters?new=1

Wizard: General → Network → Size → Verify (image, VMID, VIP).

Next / Back
Cancel
Create (starts a job)

Cluster detail

/#/clusters/:id

Header stats plus tabs: Overview, Nodes, K8s, Add-ons, Shell, Config, Upgrade, Jobs.

Download kubeconfig
Download config ZIP
Export to Terraform
Delete
Copy / View kubeconfig (Overview)

Nodes

/#/clusters/:id?tab=nodes

Guest inventory, live online/offline, CPU / memory / disk, hardware.

Add node (Create VM / Adopt / Join)
Reboot
Resize hardware
Bulk reboot / remove
Open node

Node detail

/#/clusters/:id/nodes/:nid

Inventory, Machine Health, live charts, log tail, TPM Quote.

Enroll AK
Verify Quote
Refresh / Follow logs

K8s

/#/clusters/:id?tab=k8s

Live workloads via kubectl on the mgmt host (ready clusters).

Filter namespace / kind
Refresh
Scale
Restart
Delete

Add-ons

/#/clusters/:id?tab=addons

Autoscaling, Certificates, Ingress, Storage & network.

Check config
Install / Update
Refresh

Shell

/#/clusters/:id?tab=shell

OS shell on the management host with KUBECONFIG set (operator+).

kubectl / helm in the terminal

Config

/#/clusters/:id?tab=config

Machine config YAML editor.

Apply to all nodes

Upgrade

/#/clusters/:id?tab=upgrade

Rolling Kubernetes upgrade and signed OS A/B upgrade.

Start rolling upgrade
Start OS upgrade

Jobs

/#/clusters/:id?tab=jobs

Job history with kind, status, time, duration, and color log.

Select job
Refresh
Follow

OS packages

/#/os-packages

Signed A/B OS bundles in the catalog.

Upload bundle
Upgrade clusters
Delete

Images

/#/images

Cloud qcow2 catalog (amd64 / arm64) used to create VMs.

Upload
Delete

Machines

/#/machines

All guests across clusters, live online/offline.

Filter
Refresh
Open a node

Templates

/#/templates

Reusable machine-config YAML snippets.

Create
Edit
Delete

Providers

/#/providers

Hypervisors with live CPU / memory / disk.

Add provider
Edit
Test
Delete (admin)
Open dashboard

Provider dashboard

/#/providers/:id

Hypervisor capacity gauges for the selected host / storage.

Refresh
Back to providers

Users

/#/users

Local and Auth0 identities (admin only).

Create user
Change role
Enable / disable
Send reset email

Audit

/#/audit

Action log with user, resource, and detail.

Filter action / resource
Refresh
Paging

Settings

/#/settings

Read-only session, service, paths, auth, and email config.

Switch tabs (Session / Service / Paths / Authentication / Email)

Screenshots

Dashboard — clusters and hypervisor capacity at a glance

Clusters list with status, provider, and resource bars

Cluster overview — topology, VIP, CNI, component versions

Cluster Nodes tab — inventory, reboot, resize hardware

Node detail — health, charts, logs, TPM Quote

Resize guest hardware for a control-plane node

K8s workloads — deployments with scale, restart, delete

Cluster add-ons — NFS, cert-manager, Cilium LB, Ingress, scaler

KOS scaler worker pool autoscale bounds and events

Cluster Shell — kubectl and helm on the management host

Machine config YAML applied to all nodes

Rolling Kubernetes upgrade and OS A/B upgrade

Jobs tab — kind labels, duration, and live log

Providers list — Proxmox, vSphere, Nutanix with live usage

Add provider wizard — Connection, Placement, Test

Provider dashboard — hypervisor CPU, memory, and disk

Cloud qcow2 image catalog for cluster create

Signed OS A/B packages by version and arch

Users — local accounts and Auth0 identities

Settings — session, service, paths, auth, email

pertiskd serial console dashboard on a control-plane node

pertiskd serial console dashboard on a worker node

Grafana Pertisk logs with pertiskd stream

Clone

git clone https://github.com/pertisktech/pertisk-kos.git