Browse Source

-- added k3s secure and insecure

as of now use insecure
sagnik-dev
Rick 4 months ago
parent
commit
b84f53b07e
No known key found for this signature in database GPG Key ID: 95CF60F2BBC75B7
  1. 483
      terraform/K3S/insecure.main.tf
  2. 111
      terraform/K3S/runner/nginx.yaml
  3. 112
      terraform/K3S/secure.main.tf
  4. 29
      terraform/K3S/terraform.tfvars

483
terraform/K3S/insecure.main.tf

@ -0,0 +1,483 @@
#######################################################
#### THIS IS NOT HOW YOU DEPLOY K3S IN PROD
#### THIS DOES NOT USE CERTS FOR INTERNAL COMMUNICATION
#### USE THE SECURE SCRIPT FOR ACTUAL DEPLOYMENT
####
#### By Sagnik Bhattacharya, 2024
####
#######################################################
# instaling dependency
terraform {
required_version = ">= 0.14.0"
required_providers {
openstack = {
source = "terraform-provider-openstack/openstack"
version = ">= 2.0.0"
}
tls = {
source = "hashicorp/tls"
version = ">= 3.1.0"
}
kubernetes = {
source = "hashicorp/kubernetes"
version = "~> 2.0"
}
}
}
provider "openstack" {
auth_url = var.auth_url
region = var.region
tenant_name = var.tenant_name
user_name = var.user_name
password = var.password
domain_name = var.domain_name
insecure = true # DANGER
}
variable "auth_url" {
description = "OpenStack authentication URL"
type = string
}
variable "region" {
description = "OpenStack region"
type = string
}
variable "tenant_name" {
description = "OpenStack tenant name"
type = string
}
variable "user_name" {
description = "OpenStack username"
type = string
}
variable "password" {
description = "OpenStack password"
type = string
sensitive = true
}
variable "domain_name" {
description = "OpenStack domain name"
type = string
}
# Broken for some reason dont know why
# variable "ssh_public_key" {
# description = "Path to the SSH public key"
# type = string
# }
variable "num_worker_nodes" {
description = "Number of worker nodes to create"
type = number
}
variable "master_flavor" {
description = "Flavor for the master node"
type = string
}
variable "worker_flavor" {
description = "Flavor for the worker nodes"
type = string
}
variable "os_image" {
description = "OS image to use for instances"
type = string
}
variable "volume_size" {
description = "Size of the volumes to create for nodes"
type = number
}
variable "dns_servers" {
description = "List of DNS servers for the instances"
type = list(string)
}
variable "floating_ip_pool" {
description = "Name of the floating IP pool for the instances"
type = string
}
variable "delay_seconds" {
description = "The delay in seconds before creating the worker nodes"
default = 120
## This wait time has been implimented as we were not able tp set a time till the cluster comes up
## we clould have also checked the file but in K3S as soon as the process in executed the file is created so we cant base it
## thus we habe added a timer of 120s
## depends on cluster load and netspeed
}
# Delay resource for master
resource "null_resource" "delay_master" {
provisioner "local-exec" {
command = "sleep ${var.delay_seconds}"
}
triggers = {
instance_id_master = openstack_compute_instance_v2.k3s_master.id
}
}
# Delay resource for workers
resource "null_resource" "delay_workers" {
provisioner "local-exec" {
command = "sleep ${var.delay_seconds}"
}
triggers = {
instance_id_workers = join(",", openstack_compute_instance_v2.k3s_workers.*.id)
}
}
# Define the network
resource "openstack_networking_network_v2" "network" {
name = "k3s-network"
admin_state_up = "true"
}
# Define the subnet
resource "openstack_networking_subnet_v2" "subnet" {
name = "k3s-subnet"
network_id = openstack_networking_network_v2.network.id
cidr = "192.168.1.0/24"
ip_version = 4
dns_nameservers = var.dns_servers
}
# Define the router
data "openstack_networking_network_v2" "floating_ip" {
name = var.floating_ip_pool
}
resource "openstack_networking_router_v2" "router" {
name = "k3s-router"
admin_state_up = "true"
external_network_id = data.openstack_networking_network_v2.floating_ip.id
}
# Connect the router to the subnet
resource "openstack_networking_router_interface_v2" "router_interface" {
router_id = openstack_networking_router_v2.router.id
subnet_id = openstack_networking_subnet_v2.subnet.id
}
# Adding FIP to master ## DEPRICATED
resource "openstack_networking_floatingip_v2" "fip" {
pool = var.floating_ip_pool
}
resource "openstack_compute_floatingip_associate_v2" "fip_assoc" {
floating_ip = openstack_networking_floatingip_v2.fip.address
instance_id = openstack_compute_instance_v2.k3s_master.id
}
# Creating SSH keys
resource "tls_private_key" "ssh" {
algorithm = "ECDSA"
ecdsa_curve = "P256"
}
# Saving key in local
resource "local_file" "private_key" {
content = tls_private_key.ssh.private_key_pem
filename = "${path.module}/id_rsa"
}
# Define the keypair for SSH
resource "openstack_compute_keypair_v2" "default" {
name = "k3s-key"
# public_key = file(var.ssh_public_key)
public_key = tls_private_key.ssh.public_key_openssh
}
# Create a new security group
resource "openstack_networking_secgroup_v2" "secgroup" {
name = "k3s-secgroup"
description = "Security group for k3s"
}
# # Allow SSH traffic
# resource "openstack_networking_secgroup_rule_v2" "secgroup_rule_ssh" {
# direction = "ingress"
# ethertype = "IPv4"
# protocol = "tcp"
# port_range_min = 22
# port_range_max = 22
# remote_ip_prefix = "0.0.0.0/0"
# security_group_id = openstack_networking_secgroup_v2.secgroup.id
# }
########### DONT DO THIS ITS VERY BAD ########################
# Allow all inbound traffic
resource "openstack_networking_secgroup_rule_v2" "secgroup_rule_all_inbound" {
direction = "ingress"
ethertype = "IPv4"
remote_ip_prefix = "0.0.0.0/0"
security_group_id = openstack_networking_secgroup_v2.secgroup.id
}
#############################################################
# Allow all outbound traffic
resource "openstack_networking_secgroup_rule_v2" "secgroup_rule_all_outbound" {
direction = "egress"
ethertype = "IPv4"
remote_ip_prefix = "0.0.0.0/0"
security_group_id = openstack_networking_secgroup_v2.secgroup.id
}
# Define the master node
resource "openstack_compute_instance_v2" "k3s_master" {
name = "kube-master"
image_name = var.os_image
flavor_name = var.master_flavor
key_pair = openstack_compute_keypair_v2.default.name
security_groups = ["default",openstack_networking_secgroup_v2.secgroup.name]
network {
uuid = openstack_networking_network_v2.network.id
}
# This thing does all the magic, a glorified bash script XD
user_data = <<-EOT
#!/bin/bash
apt-get update
apt-get install -y curl
echo "Before snap"
snap install helm --classic
# Install KubeCTL
curl -LO "https://dl.k8s.io/release/$(curl -L -s https://dl.k8s.io/release/stable.txt)/bin/linux/amd64/kubectl"
install -o root -g root -m 0755 kubectl /usr/local/bin/kubectl
kubectl version --client
echo "before K3S"
# Install K3s with taint if there are worker nodes
if [ ${var.num_worker_nodes} -gt 0 ]; then
curl -sfL https://get.k3s.io | sh -s - --node-taint key=value:NoExecute --disable traefik --disable-agent --tls-san 127.0.0.1
else
# Install K3s without taint, allowing the master to schedule pods
curl -sfL https://get.k3s.io | sh -s - --disable traefik --disable-agent --tls-san 127.0.0.1
fi
# Wait and save the token into a file
while [ ! -f /var/lib/rancher/k3s/server/node-token ]; do
sleep 5
done
mkdir -p /var/lib/rancher/k3s/server/
echo $(cat /var/lib/rancher/k3s/server/node-token) > /var/lib/rancher/k3s/server/token
chmod 777 /var/lib/rancher/k3s/server/token
ls -ltr /var/lib/rancher/k3s/server/token
# Mount the volume at /mnt
mkdir /mnt/data
mkfs.ext4 /dev/vdb
echo '/dev/vdb /mnt/data ext4 defaults 0 0' >> /etc/fstab
mount -a
# Adding kubeconfig
chmod 644 /etc/rancher/k3s/k3s.yaml
echo "export KUBECONFIG=/etc/rancher/k3s/k3s.yaml" >> /etc/profile
EOT
metadata = {
instance_role = "master"
}
}
# Define the volume for the master node
resource "openstack_blockstorage_volume_v3" "k3s_master_volume" {
name = "k3s-master-volume"
size = var.volume_size
}
# Attach the volume to the master node
resource "openstack_compute_volume_attach_v2" "k3s_master_volume_attach" {
instance_id = openstack_compute_instance_v2.k3s_master.id
volume_id = openstack_blockstorage_volume_v3.k3s_master_volume.id
}
resource "openstack_compute_instance_v2" "k3s_workers" {
count = var.num_worker_nodes
name = "kubeworker-${count.index}"
image_name = var.os_image
flavor_name = var.worker_flavor
key_pair = openstack_compute_keypair_v2.default.name
security_groups = ["default", openstack_networking_secgroup_v2.secgroup.name]
depends_on = [
openstack_compute_instance_v2.k3s_master,
null_resource.delay_master
]
network {
uuid = openstack_networking_network_v2.network.id
}
# This script installs necessary software and prepares the mount point
user_data = <<-EOT
#!/bin/bash
echo "hello"
apt-get update
apt-get install -y curl
# Create a mount point for the attached volume
mkdir /mnt/data
mkfs.ext4 /dev/vdb
echo '/dev/vdb /mnt/data ext4 defaults 0 0' >> /etc/fstab
mount -a
# Save the private key
echo '${tls_private_key.ssh.private_key_pem}' > /home/ubuntu/.ssh/id_rsa
chmod 600 /home/ubuntu/.ssh/id_rsa
while [ -z "$TOKEN" ]; do
TOKEN=$(ssh -o StrictHostKeyChecking=no -i /home/ubuntu/.ssh/id_rsa ubuntu@${openstack_compute_instance_v2.k3s_master.network.0.fixed_ip_v4} 'sudo cat /var/lib/rancher/k3s/server/token')
sleep 5
done
curl -sfL https://get.k3s.io | K3S_URL=https://${openstack_compute_instance_v2.k3s_master.network.0.fixed_ip_v4}:6443 K3S_TOKEN=$TOKEN sh -
EOT
# provisioner "remote-exec" {
# inline = [
# "TOKEN=$(ssh -o StrictHostKeyChecking=no -l ubuntu ${openstack_compute_instance_v2.k3s_master.network.0.fixed_ip_v4} 'cat /var/lib/rancher/k3s/server/token')",
# "curl -sfL https://get.k3s.io | K3S_URL=http://${openstack_compute_instance_v2.k3s_master.network.0.fixed_ip_v4}:6443 K3S_TOKEN=$TOKEN sh -"
# ]
connection {
type = "ssh"
user = "ubuntu"
private_key = tls_private_key.ssh.private_key_pem
host = self.access_ip_v4
}
metadata = {
instance_role = "worker"
}
}
# Define the volumes for the worker nodes
resource "openstack_blockstorage_volume_v3" "k3s_worker_volumes" {
count = var.num_worker_nodes
name = "k3s-worker-volume-${count.index}"
size = var.volume_size
}
# Attach the volumes to the worker nodes
resource "openstack_compute_volume_attach_v2" "k3s_worker_volume_attach" {
count = var.num_worker_nodes
instance_id = element(openstack_compute_instance_v2.k3s_workers.*.id, count.index)
volume_id = element(openstack_blockstorage_volume_v3.k3s_worker_volumes.*.id, count.index)
# Ensure attachment only happens after instance and volume creation
depends_on = [
openstack_compute_instance_v2.k3s_workers,
openstack_blockstorage_volume_v3.k3s_worker_volumes
]
}
## Works till here
data "kubernetes_namespace" "existing" {
metadata {
name = "kube-system"
}
}
resource "kubernetes_namespace" "default" {
count = data.kubernetes_namespace.existing.id != null ? 0 : 1
depends_on = [null_resource.delay_workers]
metadata {
name = "kube-system"
}
}
resource "kubernetes_deployment" "traefik" {
metadata {
name = "traefik"
namespace = "kube-system"
labels = {
app = "traefik"
}
}
spec {
replicas = 1
selector {
match_labels = {
app = "traefik"
}
}
template {
metadata {
labels = {
app = "traefik"
}
}
spec {
container {
name = "traefik"
image = "traefik:v2.4"
args = ["--providers.kubernetescrd", "--entrypoints.web.Address=:80", "--entrypoints.websecure.Address=:443"]
port {
name = "web"
container_port = 80
}
port {
name = "websecure"
container_port = 443
}
}
}
}
}
}
resource "kubernetes_service" "traefik" {
metadata {
name = "traefik"
namespace = "kube-system"
labels = {
app = "traefik"
}
}
spec {
selector = {
app = "traefik"
}
type = "LoadBalancer"
port {
name = "web"
port = 80
target_port = 80
}
port {
name = "websecure"
port = 443
target_port = 443
}
}
}
output "traefik_lb_ip" {
value = flatten([for s in kubernetes_service.traefik.status : [for i in s.load_balancer.ingress : i.ip]])
}

111
terraform/K3S/runner/nginx.yaml

@ -0,0 +1,111 @@
# Namespace
apiVersion: v1
kind: Namespace
metadata:
name: nginx-deployment
---
# Persistent Volume
apiVersion: v1
kind: PersistentVolume
metadata:
name: nginx-pv
spec:
capacity:
storage: 5Gi
accessModes:
- ReadWriteOnce
persistentVolumeReclaimPolicy: Retain
storageClassName: manual
hostPath:
path: "/mnt/data/nginx-pv" # Adjust this path according to the Kubernetes node filesystem
---
# Persistent Volume Claim
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: nginx-pvc
namespace: nginx-deployment
spec:
storageClassName: manual
accessModes:
- ReadWriteOnce
resources:
requests:
storage: 5Gi
---
# Deployment for Nginx Pods
apiVersion: apps/v1
kind: Deployment
metadata:
name: nginx-deployment
namespace: nginx-deployment
spec:
replicas: 2
selector:
matchLabels:
app: nginx
template:
metadata:
labels:
app: nginx
spec:
containers:
- name: nginx
image: nginx:latest
ports:
- containerPort: 80
volumeMounts:
- mountPath: /usr/share/nginx/html
name: nginx-storage
resources:
limits:
memory: "512Mi"
cpu: "1"
requests:
memory: "256Mi"
cpu: "0.5"
volumes:
- name: nginx-storage
persistentVolumeClaim:
claimName: nginx-pvc
---
# Service of type LoadBalancer to expose Nginx externally
apiVersion: v1
kind: Service
metadata:
name: nginx-service
namespace: nginx-deployment
spec:
selector:
app: nginx
ports:
- protocol: TCP
port: 80
targetPort: 80
type: LoadBalancer
---
# Ingress to route traffic to Nginx service (no specific host)
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
name: nginx-ingress
namespace: nginx-deployment
annotations:
traefik.ingress.kubernetes.io/router.entrypoints: web
spec:
rules:
- http:
# host: something.hs-fulda.de # Replace with your domain
paths:
- path: /
pathType: Prefix
backend:
service:
name: nginx-service
port:
number: 80

112
terraform/K3S/secure.main.tf

@ -0,0 +1,112 @@
#######################################################
#### Incomplete
####
#### By Sagnik Bhattacharya, 2024
####
#######################################################
terraform {
required_providers {
openstack = {
source = "terraform-provider-openstack/openstack"
version = "~> 1.0"
}
kubernetes = {
source = "hashicorp/kubernetes"
version = "~> 2.0"
}
}
}
provider "openstack" {
auth_url = var.auth_url
region = var.region
tenant_name = var.tenant_name
user_name = var.user_name
password = var.password
domain_name = var.domain_name
}
provider "kubernetes" {
host = var.kubernetes_host
client_certificate = file(var.client_certificate)
client_key = file(var.client_key)
cluster_ca_certificate = file(var.cluster_ca_certificate)
}
# Define variables without default values
variable "auth_url" {
description = "OpenStack authentication URL"
type = string
}
variable "region" {
description = "OpenStack region"
type = string
}
variable "tenant_name" {
description = "OpenStack tenant name"
type = string
}
variable "user_name" {
description = "OpenStack username"
type = string
}
variable "password" {
description = "OpenStack password"
type = string
sensitive = true
}
variable "domain_name" {
description = "OpenStack domain name"
type = string
}
variable "ssh_public_key" {
description = "Path to the SSH public key"
type = string
}
variable "kubernetes_host" {
description = "Kubernetes API server URL"
type = string
}
variable "client_certificate" {
description = "Path to the client certificate for Kubernetes"
type = string
}
variable "client_key" {
description = "Path to the client key for Kubernetes"
type = string
}
variable "cluster_ca_certificate" {
description = "Path to the cluster CA certificate for Kubernetes"
type = string
}
variable "num_worker_nodes" {
description = "Number of worker nodes to create"
type = number
}
variable "master_flavor" {
description = "Flavor for the master node"
type = string
}
variable "worker_flavor" {
description = "Flavor for the worker nodes"
type = string
}
variable "os_image" {
description = "OS image to use for instances"
type = string
}

29
terraform/K3S/terraform.tfvars

@ -0,0 +1,29 @@
## These are for connecting with Openstack and sharing the Keypair
auth_url = "https://10.32.4.182:5000/v3"
region = "RegionOne"
tenant_name = "CloudComp10" # Also known as project
user_name = "CloudComp10"
password = "demo"
domain_name = "default"
# ssh_public_key = "~/.ssh/id_ecdsa.pub"
# These are needed for the internal SSL certificate
# Must use for Pord env but for simplicity removed from here
# client_certificate = "~/.ssh/client.crt"
# client_key = "~/.ssh/client.key"
# cluster_ca_certificate = "~/.ssh/ca.crt"
# Instance Configuration
# num_worker_nodes is < 0 then master will be the worker otherwise master
# is only for control
num_worker_nodes = 3
master_flavor = "m1.small"
worker_flavor = "m1.medium"
os_image = "ubuntu-22.04-jammy-x86_64"
volume_size = 15
dns_servers = ["10.33.16.100"]
floating_ip_pool = "ext_net"
delay_seconds = 120
Loading…
Cancel
Save