01
Device edge & far edge
Industrial gateways, sensor aggregators, remote monitoring nodes. Full Kubernetes where connectivity is intermittent and RAM is measured in megabytes.
edgeopen source · MIT · by portainer
Kubernetes for everywhere
you run containers.
Anywhere you've been running Docker or Podman, you can now run Kubernetes. KubeSolo strips out all the clustering machinery and gives you the full Kubernetes API on a single node… starting at under 200 MB RAM.
curl -sfL https://get.kubesolo.io | sudo sh -
MEMORY
<200 MB
RAM at idle
PLATFORMS
4 arch
ARM · ARM64 · x86_64 · RISC-V 64
OVERHEAD
Zero
Cluster overhead
API STATUS
Full K8s
API compatible
Most Kubernetes distributions are built for multi-node clusters and then adapted for smaller deployments. KubeSolo starts from the opposite direction. It takes Kubernetes and removes everything that only makes sense when there's more than one node: etcd quorum logic, leader election, multi-node networking overlays, control plane distribution. None of it is present.
What remains is a full Kubernetes control loop on a single process. The API server, scheduler, controller manager, and kubelet all run together. Your existing manifests, Helm charts, and CRDs work without modification.
The design target is anywhere you would have previously reached for Docker or Podman: edge devices, factory hardware, developer laptops, remote sites, IoT gateways, kiosk machines. Anywhere a single-node runtime makes sense and a full cluster is overhead you don't need.
KubeSolo is open source (MIT) and maintained by Portainer.
1# install KubeSolo
2curl -sfL https://get.kubesolo.io | sudo sh -
4[INFO] Detecting architecture... arm64
5[INFO] Downloading kubesolo v1.0.0
6[OK] Service started successfully
8# set kubeconfig context
9$ kubectl config use-context kubernetes-admin@kubesolo
10$ kubectl get nodes
11rpi-gate Ready <none> 8s v1.30.0
13$ kubectl apply -f mosquitto.yaml
14namespace/mosquitto created
02
Factory floors & OT
Air-gapped industrial infrastructure running ISA-95 workloads. KubeSolo runs fully offline with no dependency on cloud connectivity or external registries.
air-gapped03
IoT & embedded systems
Raspberry Pis, ARM SBCs, embedded Linux devices that were previously running Docker Compose. Upgrade the runtime without upgrading the hardware.
arm6404
Developer workstations
Local Kubernetes that doesn't need a hypervisor or eat your laptop's memory. Write manifests locally, deploy identically to production. No Docker Desktop required.
local dev05
Remote & branch sites
Small offices, retail locations, field deployments. One KubeSolo node per site with the same operational model as the rest of your fleet.
fleet06
Kiosk & appliance workloads
Interactive terminals, digital signage, point-of-sale systems. Lifecycle-managed container workloads on locked-down purpose-built hardware.
embedded✗No declarative desired-state management
✗No built-in RBAC or multi-tenant access control
✗No Helm or CRD support
✗No native health management or automatic restarts beyond basic flags
✗No standard ecosystem tooling (operators, service mesh, GitOps)
✗Compose files don't translate to production K8s deployments
✓Declarative workload management via standard manifests
✓Full RBAC, ServiceAccounts, and namespace isolation
✓Helm and CRD support out of the box
✓Native liveness/readiness probes, restart policies, resource limits
✓Full ecosystem compatibility: operators, Flux, ArgoCD, Prometheus
✓Same OCI images, no changes to your existing container builds
| Capability | KubeSolo | K3s | MicroK8s | Docker / Podman |
|---|---|---|---|---|
| Full Kubernetes API | ✓ | ✓ | ✓ | ✗ |
| RAM footprint <200 MB | ✓ | ~ | ✗ | ✓ |
| Single-node design (no cluster overhead) | ✓ | ~ | ~ | ✓ |
| ARM, ARM64, x86_64 & RISC-V 64 | ✓ | ✓ | ✓ | ✓ |
| Flash / constrained storage optimized | ✓ | ~ | ✗ | ~ |
| Helm & CRD support | ✓ | ✓ | ✓ | ✗ |
| No etcd / no quorum | ✓ | ✓ | ✗ | ✓ |
| Air-gapped install | ✓ | ✓ | ~ | ✓ |
| OCI image compatible | ✓ | ✓ | ✓ | ✓ |
1
Install
Run the install script with sudo. This installs the KubeSolo binary and starts the control plane as a systemd service. Ensure no existing container engine (Docker, Podman, containerd) is running first.
BASH
1$ curl -sfL https://get.kubesolo.io | sudo sh -
2
Set up kubectl
Copy the admin kubeconfig from /var/lib/kubesolo/pki/admin/admin.kubeconfig to the machine where kubectl is installed, then set the context:
BASH
1$ kubectl config use-context kubernetes-admin@kubesolo
2$ kubectl get nodes
3# You should see a single node in Ready state
3
Deploy your first workload
Try the Mosquitto example from the KubeSolo repository:
BASH
1$ kubectl apply -f https://raw.githubusercontent.com/portainer/kubesolo/develop/examples/mosquitto.yaml
2$ kubectl get all -n mosquitto
Managing multiple nodes? Add Portainer.
KubeSolo is a Portainer project. Portainer's Edge Agent connects your KubeSolo fleet into a single operator control plane: GitOps deployments, RBAC, and remote lifecycle management across every node without touching each one.