NodePort vs ClusterIP

kubectl expose deployment nginx --port=80 --type=NodePort

kubectl expose deployment nginx --port=80 --type=ClusterIP

ClusterIP (Default)

kubectl expose deployment nginx \
    --port=80 \
    --type=ClusterIP

Accessible from

✅ Pods inside the cluster

❌ Host machine

❌ Your laptop

❌ Internet

Architecture:

          Pod A
             │
             │
      nginx-service
      10.96.85.101
             │
             │
          nginx Pod

The service gets a virtual IP (ClusterIP).

Example:

kubectl get svc

NAME            TYPE        CLUSTER-IP      PORT
nginx-service   ClusterIP   10.96.85.101    80

Pods communicate using

http://nginx-service

or

http://10.96.85.101

NodePort

kubectl expose deployment nginx \
    --port=80 \
    --type=NodePort

Accessible from

✅ Pods

✅ Kubernetes nodes

✅ Host

✅ Other computers (if routing/firewall permits)

Architecture

Browser
   │
   │
192.168.100.10:32180
   │
NodePort Service
   │
   │
nginx Pod

Example

kubectl get svc

NAME            TYPE       CLUSTER-IP      PORT(S)

nginx-service   NodePort   10.96.85.101    80:32180/TCP

Here

32180

is automatically allocated.

Access it:

http://192.168.100.10:32180

---

Visual Comparison

ClusterIP

           Internet
               X

           Host
               X

----------------------------

     Kubernetes Cluster

 Pod ---- Service ---- Pod

Only internal communication.

---

NodePort

Internet
     │
     │
Host
     │
     │
Node:32180
     │
     │
Service
     │
     │
Pods

The service is exposed on every Kubernetes node.

---

In your LXD lab

Suppose

k8master
192.168.100.10

k8worker1
192.168.100.11

NodePort

80:32180

You can access

http://192.168.100.10:32180

or

http://192.168.100.11:32180

Create LXD in Ubunut 26 to train kubernetes from scratch

Create a preseed for initializing LXD (storage pool, bridge, default profile).

• Ubuntu 26.04 (latest)

• LXD 5.21 LTS (Snap) (latest)

• Kubernetes 1.33.x (latest)

Storage Pool

-------------

default

Bridge

------

k8sbr0

Profile

-------

k8s-base

Containers

----------

k8master

k8worker1

we'll build LXD properly from scratch with:

• Clean LXD initialization.

• ZFS storage pool.

• k8sbr0 bridge.

• DHCP reservations.

• Static container IPs.

• The reusable k8s-base profile.

• Creation of k8master and k8worker1.

• One reusable profile (k8s-base).

• One bridge (k8sbr0).

• One storage pool (default).

• One common node installation script.

• One rebuild script.

• One validation script.

• Snapshots after each major milestone.