Kalisto/ documentation source (permission required) ↗
On this page

#Workloads, add-ons and ingress

Start with an explicitly prepared Cilium profile. After successful up, Cilium and CoreDNS provide pod networking/DNS. Ingress is not installed by default.

#Apply your manifests

With a host Kubernetes client:

export KUBECONFIG="$(kalisto kubeconfig --path)"
kubectl get nodes
kubectl -n kube-system get pods
kubectl apply -f app.yaml

Or without host kubectl:

kalisto apply --name kalisto --manifest app.yaml
kalisto kubectl --name kalisto -- get nodes

app.yaml is user-provided. Managed kubectl is checksum-verified and matches the cluster's K3s release; pull prepares it, otherwise first use can fetch it, not up. No standalone kubectl command is added to PATH. On shared hosts, pass the explicit cluster name when selecting kubeconfig. Workload images can need registry access: the startup cache does not contain every application.

#Install ingress after prepared startup

These are explicit online operations, outside the prepared boot budget. Prepared startup only supports no CNI or Cilium and optional lb/servicelb; it rejects other startup add-ons instead of downloading them silently.

# Traefik ingress + MetalLB
kalisto addon install --name kalisto ingress/traefik
kalisto addon install --name kalisto lb/metallb

# Or Kong ingress + MetalLB
kalisto addon install --name kalisto ingress/kong
kalisto addon install --name kalisto lb/metallb

#MetalLB configuration

Configure addresses only through kalisto.toml. Add this after prepared base startup; enabling it before pull selects an unsupported offline profile:

[plugins.metallb]
enabled = true
mode = "layer2"
addresses = ["192.168.1.240-192.168.1.250"]

Alternatively provide a full manifest from disk:

[plugins.metallb]
enabled = true
config_file = "manifests/metallb.yaml"

addresses and config_file are mutually exclusive. Setting either auto-enables lb/metallb. Select ranges reachable from your host network that do not overlap active DHCP/static assignments. Example addresses are not a recommendation to use that range on your network.

#Ingress defaults

The checked-in add-on manifests configure:

  • Traefik: service.type=LoadBalancer; Kubernetes Ingress publishedService enabled.
  • Kong: ingressClass=kong, createIngressClass=true, ingressController.env.publish_service=kube-system/kong-kong-proxy, proxy.type=LoadBalancer.

Add-on install does not apply a global domain annotation by default.

#Ingress hostnames

Create an explicit Kubernetes Ingress and set spec.rules[].host to your domain or an IP-based hostname from sslip.io. Discover the selected controller's LoadBalancer IP:

KUBECONFIG_PATH="$(kalisto kubeconfig --path)"
kubectl --kubeconfig "$KUBECONFIG_PATH" -n kube-system get svc traefik
kubectl --kubeconfig "$KUBECONFIG_PATH" -n kube-system get svc kong-kong-proxy

Use myapp.<LB_IP>.sslip.io, for example myapp.192.168.1.50.sslip.io, replacing the IP with the actual reachable service address.

For an existing Ingress, replace the old hostname in spec.rules[].host and reapply its manifest. If HTTPS is configured, update the matching spec.tls[].hosts entries and provision a certificate covering the new hostname through your existing certificate workflow; an old hostname's certificate will not match.

This service supplies DNS only: it does not install an ingress controller, proxy traffic, open ports, make a private address reachable, or issue TLS certificates. The IP must already be reachable from the client. Some resolvers block private or loopback answers as DNS-rebinding protection; use an explicit local DNS record or hosts entry for the same Ingress hostname when necessary, rather than disabling protection globally.

#Traefik example

apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: myapp-traefik
spec:
  ingressClassName: traefik
  rules:
    - host: myapp.192.168.1.50.sslip.io
      http:
        paths:
          - path: /
            pathType: Prefix
            backend:
              service:
                name: myapp
                port:
                  number: 80

#Kong example

apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: myapp-kong
spec:
  ingressClassName: kong
  rules:
    - host: myapp.192.168.1.50.sslip.io
      http:
        paths:
          - path: /
            pathType: Prefix
            backend:
              service:
                name: myapp
                port:
                  number: 80

#Planned expose command—not implemented

The proposed target UX is:

# Planned only: this command is not implemented.
kalisto expose svc myapp --port 80 --ingress traefik

The intended behavior is to discover the ingress LoadBalancer IP, generate <service>.<lb-ip>.sslip.io, create/update an Ingress with the selected class, optionally annotate the controller service for domain automation, and print the final URL. None of this is an implemented kalisto expose svc workflow. Use the explicit manifests above. MetalLB IP pool configuration remains config-only in kalisto.toml, and add-on installation does not add a global domain annotation.