The Value of Network Administration Skills in Cloud Computing

Seun BSeun B
33 min read

In the cloud computing era, network administration skills are invaluable. Hybrid and multi-cloud environments require traditional networking knowledge to integrate on-premises infrastructure with cloud services. Cloud networking concepts like VLANs, subnets, routing, and firewalls are essential.

Network security skills apply directly to cloud security measures. Understanding containerization, microservices, and Infrastructure as Code (IaC) is crucial. Network troubleshooting, edge computing, compliance, cost optimization, and cloud-native networking all benefit from a strong foundation in traditional networking principles.

Lets picture a scenario where an organization is opening a new branch office and needs to set up a secure network infrastructure that integrates with the main office. As a Cloud Platform Engineer, you’re tasked with designing, implementing, and automating this setup. I will be running through a step by step process of how to achieve this using a vmware worksation. The concept remains the same whether you use on-prem or cloud resources.

Deploying 3 VMs Using VMware Workstation

Network Architecture Diagram:

A Simple Network Architecture

Diagram Details:

  • Main Office VPN Gateway: Connects to the Branch Router via a VPN tunnel.

  • Branch Router: Provides internet access and routes traffic between the branch office and the main office via the VPN tunnel.

  • Branch Switch: Central hub connecting all devices within the branch office.

  • Peripheral Devices: Connected directly to the Branch Switch.

  • Branch Server: Manages DNS, VPN, DHCP, and NTP services.

  • Client Desktops: Connect to the Branch Switch to access network resources and services.

Using VMware Workstation, I took the steps below to deploy 3 VMs and achieve the network configuration:

Create Three VMs

Create three VMs using VMware Workstation and a predefined VMDK disk (located at OSBoxes),https://www.osboxes.org/ubuntu/#ubuntu-24-04-vmware follow these steps:

  1. Open VMware Workstation

  2. Start VMware Workstation on your machine.

  3. Create a New Virtual Machine

  4. For each of the three VMs (Main Office Server, Branch Office Server, and Client VM), follow these steps:

  • Select File > New Virtual Machine…

  • Specify Disk File

  • Select I will install the operating system later and click next.

  • Select a guest operating system Linux and version ubuntu

  • Choose the location where you have kept the custom .vmdk file Complete the setup and click finish

  • Name the Virtual Machine

  • Give a name to the virtual machine, e.g., “Main Office Server,” “Branch Office Server,” or “Client VM.” Specify the location where you want to store the VM files.

  • Finish

  • Click “Finish” to complete the VM creation process.

  • Repeat these steps for each of the three VMs.

    • Now select a virtual machine, click on VM then settings, this opens the VM properties, click Add, select hard disk, scsi, use an existing virtual disk, browse for location of disk and click finish.

Now you can start the VM with the imported disk [.vmdk file] you have selected.

Network Configuration

Create Virtual Networks

  1. Create Virtual Networks in VMware Workstation:

MainOfficeNet:

  1. Open VMware Workstation.

  2. Go to Edit > Virtual Network Editor.

  3. Click Add Network and select a network (e.g., VMnet2).

  4. Set the network to Host-only.

  5. Set the subnet IP to 192.168.1.0 and the subnet mask to 255.255.255.0.

  6. Disable the DHCP for this network.

  7. Rename VMnet3 to MainOfficeNet.

  • BranchOfficeNet:
  1. Follow the same steps to create another network (e.g., VMnet3).

  2. Set the network to Host-only.

  3. Set the subnet IP to 192.168.5.0 and the subnet mask to 255.255.255.0.

  4. Disable the DHCP for this network.

  5. Rename VMnet3 to BranchOfficeNet.

  1. Assign Networks to VMs:
  • Main Office Server:
  1. Go to the settings of the Main Office Server VM.

  2. Under Network Adapter, add the network adapter.

  3. Attach the adapter to VMnet2 (MainOfficeNet).

  • Branch Office Server:
  1. Go to the settings of the Branch Office Server VM.

  2. Under Network Adapter, add the network adapters.

  3. Attach the adapter to VMnet3 (BranchOfficeNet).

  • Client VM:
  1. Go to the settings of the Client VM.

  2. Under Network Adapter, attach the adapter to VMnet3 (BranchOfficeNet).

  3. Power On the VMs

  4. Power on each VM by right-clicking each VM and selecting “Power On.”

Configure Network Interfaces

Updating or Replacing Existing Connections

  1. Identify and Modify Existing Connections
  • Delete Existing Connections:
sudo nmcli con delete netplan-ens33 
sudo nmcli con delete 'Wired connection 1'
  • Add New Connections:

  • Add MainOfficeNet to ens33

sudo nmcli con add type ethernet ifname ens33 con-name MainOfficeNet ip4 192.168.1.10/24
  1. Apply and Bring Up Connections
  • Bring Up the Connection:

  • sudo nmcli con up MainOfficeNet

  1. Restart NetworkManager

  2. Restart NetworkManager to apply the changes:

sudo systemctl restart NetworkManager

Configuration for Branch Office Server

Configure Network Interfaces:

  • Delete Existing Connections:
sudo nmcli con delete netplan-ens33 
sudo nmcli con delete 'Wired connection 1'
  • Add New Connections:

  • # Add BranchOfficeNet to ens33

sudo nmcli con add type ethernet ifname ens33 con-name BranchOfficeNet ip4 192.168.5.10/24
  1. Bring Up the Connections:
  • sudo nmcli con up BranchOfficeNet
  1. Verify the configuration:
nmcli con show 
nmcli con show --active 
ip addr show 
ifconfig -a

Configuration for Client VM

  1. Delete Existing Connections:
  • sudo nmcli con delete 'Wired connection 1'

Add DHCP Configuration:

  • nmcli con add type ethernet ifname ens33 con-name BranchOfficeNet ipv4.method auto

Bring Up the Connection:

  • sudo nmcli con up BranchOfficeNet

Verify Configuration

  • From the Client VM, ensure it gets an IP address via DHCP and check connectivity:

  • nmcli device show ens33 ip route ip addr

Setting Up DNS:

Setting up DNS with BIND9 involves configuring both the main and branch servers to handle local domain resolution and forward external queries appropriately. Below are the detailed steps for setting this up.

Main Server (main.abc.local)

1. Install BIND9

sudo apt update
sudo apt install bind9 bind9utils bind9-doc -y

2. Configure BIND9

Edit the BIND9 configuration files to set up the master server:

  • Edit named.conf.local

  • sudo nano /etc/bind/named.conf.local

Add the following:

zone "abc.local" {
    type master;
    file "/etc/bind/db.abc.local";
};

zone "5.168.192.in-addr.arpa" {
    type master;
    file "/etc/bind/db.192.168.5";
};

Create Zone Files

  • Create the forward zone file for abc.local:

  • sudo nano /etc/bind/db.abc.local

Add the following content:

$TTL    604800
@       IN      SOA     main.abc.local. admin.abc.local. (
                             2         ; Serial
                        604800         ; Refresh
                         86400         ; Retry
                       2419200         ; Expire
                        604800 )       ; Negative Cache TTL
;
@       IN      NS      main.abc.local.
main    IN      A       192.168.1.10
branch  IN      A       192.168.5.10
client  IN      A       192.168.5.15
  • Create the reverse zone file for 192.168.5.x:

  • sudo nano /etc/bind/db.192.168.5

  • Add the following content:

$TTL    604800
@       IN      SOA     main.abc.local. admin.abc.local. (
                             2         ; Serial
                        604800         ; Refresh
                         86400         ; Retry
                       2419200         ; Expire
                        604800 )       ; Negative Cache TTL
;
@       IN      NS      main.abc.local.
10      IN      PTR     branch.abc.local.
15      IN      PTR     client.abc.local.

3. Configure Forwarders

Edit named.conf.options to include forwarders:

sudo nano /etc/bind/named.conf.options

Add the following within the options block:

options {
    directory "/var/cache/bind";

    forwarders {
        8.8.8.8;  // Google's DNS
        8.8.4.4;  // Google's DNS
    };

    dnssec-validation auto;

    listen-on-v6 { any; };
};

4. Restart BIND9

sudo systemctl restart bind9
systemctl status bind9
systemctl restart named.service

Branch Server (branch.abc.local)

1. Install BIND9

sudo apt update
sudo apt install bind9 bind9utils bind9-doc -y

Configure iptables to permit incoming dns queries from client systems.

Allow incoming DNS traffic on port 53 (UDP)

sudo iptables -A INPUT -p udp - dport 53 -j ACCEPT

Allow incoming DNS traffic on port 53 (TCP)

sudo iptables -A INPUT -p tcp - dport 53 -j ACCEPT

Save the iptables rules

sudo sh -c “iptables-save > /etc/iptables/rules.v4”

2. Configure BIND9 as Slave

Edit the BIND9 configuration files to set up the branch server:

  • Edit named.conf.local

  • sudo nano /etc/bind/named.conf.local

  • Add the following:

zone "abc.local" {
    type slave;
    file "/var/cache/bind/db.abc.local";
    masters { 192.168.1.10; };
};

zone "5.168.192.in-addr.arpa" {
    type slave;
    file "/var/cache/bind/db.192.168.5";
    masters { 192.168.1.10; };
};
  • Configure Forwarding

  • Edit named.conf.options to forward unknown/external queries to the main server:

  • sudo nano /etc/bind/named.conf.options

  • Add the following within the options block:

options {
    directory "/var/cache/bind";

    forwarders {
        192.168.1.10;  // Main server's IP
    };

    dnssec-validation auto;

    listen-on-v6 { any; };
};

3. Restart and enable BIND9 service at every system reboot

sudo systemctl restart bind9
sudo systemctl enable bind9

Check Zone Transfer Logs on Branch Server

sudo systemctl status bind9

Test DNS Resolution

On the branch server, use dig to test:

dig main.abc.local @localhost
dig branch.abc.local @localhost
dig client.abc.local @localhost
dig google.com @localhost

The dig output indicates that the DNS server running on localhost successfully resolved the domain main.abc.local to the IP address 192.168.1.10, but it issued a warning because .local is reserved for mDNS. The query was handled correctly with an authoritative answer and the response took 4 milliseconds.

The dig output indicates that the DNS server running on localhost successfully resolved google.com to the IP address 142.251.32.78. The query was handled correctly with an authoritative answer, and the response took 1 millisecond. The result shows that your local DNS server is capable of resolving external domain names as well, as it is properly configured to forward queries or perform DNS resolution.

By following these steps, your branch server will be configured to resolve local domain queries from its own zone files and forward any unknown or external domain queries to the main server. The main server is configured to forward external queries to public DNS servers (like Google’s DNS).

Client Configuration (192.168.5.15)

Ensure that the client machine is configured to use the branch office DNS server for its DNS queries.

Setup # operation for /etc/resolv.conf.

nameserver 192.168.5.10 nameserver 127.0.0.53 options edns0 trust-ad search branch.company.local

  1. Configure/etc/netplan/01-netcfg.yaml

  2. Edit the netplan configuration to use the branch server for DNS:

  • sudo nano /etc/netplan/01-netcfg.yaml
  1. Ensure it contains:

network:
  version: 2
  ethernets:
    ens33:
      dhcp4: yes
      routes:
        - to: default
          via: 192.168.5.10
      nameservers:
        addresses:
          - 192.168.5.10
          - 8.8.8.8

Apply the Netplan configuration:

sudo netplan apply

Verify the Setup

  1. Check DNS Resolution

  2. On the client, verify that DNS resolution works:

ping main.abc.local 
ping branch.abc.local  
nslookup main.abc.local 
nslookup branch.abc.local  
nslookup client.abc.local

External DNS Resolution:

  • nslookup google.com

  1. Test External DNS Resolution

  2. On the branch server, test that external domain resolution works:

  • dig google.com
  1. Check BIND9 Status

  2. Ensure that BIND9 is running correctly on both the main and branch servers:

  • sudo systemctl status bind9

By following these steps, you should have a working DNS setup where the branch office can resolve both local and external domains through the main office DNS server.

Steps to Deploy Tinc VPN

Tinc VPN is a flexible and powerful VPN daemon that supports full-mesh routing and dynamic links between nodes. Here’s how to deploy Tinc VPN on both the main server and branch server.

Prerequisites

  • Two Linux servers (main and branch) with root or sudo access.

  • Ensure that both servers have open network ports for Tinc (default is 655 for TCP and UDP).

Step 1: Install Tinc VPN

On Both Servers [main and branch]:

  1. Update the package list and install Tinc:
  • sudo apt update sudo apt install tinc

Step 2: Create Tinc Configuration Directories

On Both Servers:

  1. Create the main configuration directory for Tinc:
  • sudo mkdir -p /etc/tinc/vpn/hosts
  1. Navigate to the Tinc directory:
  • cd /etc/tinc/vpn

Step 3: Generate Tinc Configuration Files

On Both Servers:

  1. Create thetinc.conffile:
  • sudo nano tinc.conf
  1. Add the following configuration (replaceMainServerandBranchServerwith appropriate hostnames):

  2. Main Server (main):

Name = main  
AddressFamily = ipv4  
Interface = tun0

  1. Branch Server (branch):
Name = branch  
AddressFamily = ipv4  
Interface = tun0  
ConnectTo = main
  1. Create thetinc-upscript:
  • sudo nano tinc-up
  1. Add the following content (adjust IP addresses as needed):

  2. Main Server:

  • #!/bin/sh ifconfig $INTERFACE 10.0.0.1 netmask 255.255.255.0

  1. Branch Server:
  • #!/bin/sh ifconfig $INTERFACE 10.0.0.2 netmask 255.255.255.0
  1. Make thetinc-upscript executable:
  • sudo chmod +x tinc-up
  1. Create thetinc-downscript:
  • sudo nano tinc-down
  1. Add the following content:
  • #!/bin/sh ifconfig $INTERFACE down
  1. Make thetinc-downscript executable:
  • sudo chmod +x tinc-down

Step 4: Configure Host Files

On Both Servers:

  1. Create a host configuration file for each server:

  2. Main Server:

  • sudo nano hosts/main
  1. Branch Server:
  • sudo nano hosts/branch
  1. Add the following content:

Main Server: [hosts/main]

  • Address = 192.168.1.10 Subnet = 10.0.0.1/32

Branch Server: [hosts/branch]

  • Address = 192.168.5.10 Subnet = 10.0.0.2/32

Step 5: Generate Tinc Keys

On Both Servers:

  1. Generate the Tinc RSA key pair:
sudo tincd -n vpn -K4096

This will generatersa_key.privandhosts/<hostname>files.
Share the contents of these files between the servers:

Main Server:

  • sudo cat /etc/tinc/vpn/hosts/main

Branch Server:

  • sudo cat /etc/tinc/vpn/hosts/branch
  1. Copy the public key portion (the lines starting with-----BEGIN RSA PUBLIC KEY-----to-----END RSA PUBLIC KEY-----) from each server and add it to the corresponding host file on the other server:

  2. On Main Server (/etc/tinc/vpn/hosts/branch):

  • -----BEGIN RSA PUBLIC KEY----- (Branch Server public key here) -----END RSA PUBLIC KEY-----
  1. On Branch Server (/etc/tinc/vpn/hosts/main):
  • -----BEGIN RSA PUBLIC KEY----- (Main Server public key here) -----END RSA PUBLIC KEY-----

  1. Ensure the hosts files, main and branch exists on both servers. [copy them all to each other]

Step 6: Start Tinc VPN

On Both Servers:

  1. Enable and start the Tinc service:
  • sudo systemctl enable tinc@vpn sudo systemctl start tinc@vpn sudo systemctl status tinc@vpn
  1. Check the status of the Tinc service:
  • sudo systemctl status tinc@vpn

Step 7: Verify the Connection

  1. Verify that thetun0interface is up and configured correctly on both servers:
  • ifconfig tun0

  1. Check the connectivity between the servers:
  • ping 10.0.0.2 # From Main Server ping 10.0.0.1 # From Branch Server ssh osboxes@10.0.0.2 # From Main Server ifconfig tun0

Setup DHCP on Branch Office:

  1. Install DHCP (isc-dhcp-server):
  • sudo apt update sudo apt install isc-dhcp-server
  1. Configure DHCP Server:

  2. Edit the DHCP server configuration file:

  • sudo nano /etc/dhcp/dhcpd.conf
  1. Add or modify the following lines to configure the DHCP server:
# Optionally specify a domain name
option domain-name "company.local";

# Specify the default lease time (in seconds)
default-lease-time 600;

# Specify the maximum lease time (in seconds)
max-lease-time 7200;

# Specify the network and subnet
subnet 192.168.5.0 netmask 255.255.255.0 {
    range 192.168.5.50 192.168.5.100;  # IP range to be assigned to clients
    option routers 192.168.5.10;       # Gateway
    option domain-name-servers 192.168.5.10, 8.8.8.8;  # DNS servers

  1. Specify Network Interface for DHCP:

  2. Edit the file /etc/default/isc-dhcp-server to specify the network interface that the DHCP server should listen on. For example:

INTERFACESv4="ens33"

Replace ens33 with the name of the network interface connected to your local network.

  1. Restart DHCP Server:

  2. Activate and start the DHCP server to apply the changes:

sudo systemctl start isc-dhcp-server
systemctl status isc-dhcp-server
systemctl enable isc-dhcp-server

On the Client Server

1. Configure Netplan for DHCP

Edit the Netplan configuration file to use DHCP for obtaining an IP address:

  1. Edit Netplan Configuration:
sudo nano /etc/netplan/01-netcfg.yaml
  1. Modify the Configuration to Use DHCP:

  2. Update the file to use DHCP for the ens33 interface:

network:
version: 2
ethernets:
 ens33:
   dhcp4: yes
   routes:
     - to: default
       via: 192.168.5.10
   nameservers:
     addresses:
       - 8.8.8.8
       - 8.8.4.4

  1. Apply the Netplan configuration:
  • sudo netplan apply
  1. A new network will be created and will then need to be attached to ens33, then reboot so that it can pickup an IP.

  1. Or you can use below command to configure the network to be auto assigned an IP.
sudo nmcli con add type ethernet ifname ens33 con-name netplan-ens33 ipv4.method auto

Verify the Setup

  1. Check DHCP Lease on Client:
nmcli con show 
sudo nmcli con up netplan-ens33 
nmcli device show ens33 
nmcli device status 
nmcli con show netplan-ens33

Once the client server is powered up, it should automatically obtain an IP address from the branch server. Verify the IP address on the client:

ip addr show ens33

You should see an IP address within the range specified in the DHCP server configuration (e.g., 192.168.5.15 to 192.168.5.40).

By following these steps, your branch server will act as a DHCP server, and the client server will automatically receive its IP address from the branch server when powered up.

Provide Internet Access from Branch Office Server to Client System

To provide internet access from your branch office server (which has internet access) to your client system (which does not), you can set up Network Address Translation (NAT) and configure IP forwarding on the branch office server. Here’s a step-by-step guide to achieve this on your Ubuntu server:

1. Enable IP Forwarding

You need to enable IP forwarding on the branch office server to allow it to forward packets between the client system and the internet.

Open the /etc/sysctl.conf file with a text editor:

sudo nano /etc/sysctl.conf

Find the line:

#net.ipv4.ip_forward=1

Uncomment it (remove the #), so it reads:

net.ipv4.ip_forward=1

Apply the changes:

sudo sysctl -p

2. Configure NAT (Network Address Translation)

Use iptables to configure NAT on the branch office server. This will allow the client system to use the branch office server’s internet connection.

Run the following commands:

sudo iptables -t nat -A POSTROUTING -o ens38 -j MASQUERADE
sudo iptables -A FORWARD -i ens38 -o ens33 -m state --state RELATED,ESTABLISHED -j ACCEPT
sudo iptables -A FORWARD -i ens33 -o ens38 -j ACCEPT

Replace <internet_interface> with the name of the network interface connected to the internet (e.g., ens33) and <client_interface> with the name of the network interface [BranchOfficeNet]connected to the client system (e.g., ens34).

3. Save theiptables Configuration

To ensure that your iptables rules persist after a reboot, you need to save them. On Ubuntu, you can use the iptables-save command and store the rules in a file.

Save the rules:

Install iptables-persistent to load the rules at boot [if not already installed]:

sudo apt-get install iptables-persistent
sudo sh -c "iptables-save > /etc/iptables/rules.v4"
OR
sudo iptables-save | sudo tee /etc/iptables/rules.v4

Check that the NAT rule is in place:

sudo iptables -t nat -L -v

4. Configure the Client System

Ensure that the client system has its default gateway set to the branch office server’s IP address on the local network.

On the client system, you can set the default gateway by editing the /etc/netplan/01-netcfg.yaml file or using the ip route command. For example:

sudo nano /etc/netplan/01-netcfg.yaml
network:
version: 2
ethernets: ens33:
dhcp4: yes
routes:

network:
  version: 2
  ethernets:
    ens33:
      dhcp4: yes
      routes:
        - to: default
          via: 192.168.5.10
      nameservers:
        addresses:
          - 192.168.5.10
          - 8.8.8.8
sudo netplan apply

5. Test the Configuration

Test the internet connectivity from the client system by pinging an external website or using a web browser.

ping google.com

NTP setup

I will be using chrony set up the NTP service, where the client system receives time sychronization from the branch server, following these steps:

On the Branch Server

  1. Install Chrony:
sudo apt-get update
sudo apt-get install chrony -y
  1. Configure Chrony:

Edit the Chrony configuration file:

sudo nano /etc/chrony/chrony.conf

Add the following lines to allow the client server to get time from the branch server:

allow 192.168.5.0/24

This allow directive permits the specified network to access the time service. Adjust the network range if necessary.

Configure IPtables to permit NTP service syncronization on port 123 from client systems:

sudo iptables -A INPUT -p udp --dport 123 -j ACCEPT
sudo iptables -A OUTPUT -p udp --sport 123 -j ACCEPT
sudo iptables-save | sudo tee /etc/iptables/rules.v4
sudo iptables -L -v -n

  1. Start Chrony:
sudo systemctl start chrony
sudo systemctl enable chrony
  1. Verify Chrony Status:
sudo systemctl status chrony

On the Client Server

  1. Install Chrony:
sudo apt-get update
sudo apt-get install chrony -y
  1. Configure Chrony:

Edit the Chrony configuration file:

sudo nano /etc/chrony/chrony.conf

Add the branch server’s IP address as the NTP server, remove all other default ntp sources by hashing it out:

server 192.168.5.10 iburst #Branch server IP address

  1. Start Chrony:
sudo systemctl start chrony
  1. Verify Chrony Status:
sudo systemctl status chrony

Verification

On the brannch server, run ‘netplan apply’

To ensure that the client server is correctly synchronizing its time from the branch server, you can use the chronyc command.

On the Client Server:

  1. Check Chrony Sources:
chronyc sources

You should see the branch server (192.168.5.10 or its hostname) listed as a source.

  1. Other chrony commands are listed below:
chronyc tracking
sudo chronyc -a makestep
chronyc activity
chronyc serverstats
chronyc sources -v

To set up Snort to monitor for suspicious activity on your servers, follow these steps:

1. Install Snort

On Debian/Ubuntu:

sudo apt update
sudo apt install snort

2. Configure Snort

Initial Configuration:

Snort’s main configuration file is located at /etc/snort/snort.conf. Open this file to configure it according to your network setup.

sudo nano /etc/snort/snort.conf

Configure Network Variables:

Set your HOME_NET and EXTERNAL_NET variables. set up Snort to monitor three interfaces with different subnet, set it as follows:

for Main server:

var HOME_NET [192.168.1.0/24,10.0.0.0/24,192.168.79.0/24]
var EXTERNAL_NET !$HOME_NET

for branch server:

var HOME_NET [192.168.5.0/24,10.0.0.0/24,192.168.79.0/24] 
var EXTERNAL_NET !$HOME_NET

for client:

var HOME_NET 192.168.5.0/24 
var EXTERNAL_NET !$HOME_NET

Include Rule Files:

Ensure the rule paths are correctly specified:

echo 'include $RULE_PATH/local.rules' >> /etc/snort/snort.conf

4. Create Local Rules

You can create custom rules specific to your network in the local.rules file:

sudo nano /etc/snort/rules/local.rules

Add some basic rules:

# Alert on any ICMP traffic
alert icmp any any -> any any (msg:"ICMP Traffic Detected"; sid:1000001; rev:1;)

# Alert on any TCP traffic to port 123 (NTP)
alert tcp any any -> any 123 (msg:"TCP Traffic to NTP port 123 detected"; sid:1000002; rev:1;)

# Alert on any TCP traffic to port 53 (DNS)
alert tcp any any -> any 53 (msg:"TCP Traffic to DNS port 53 detected"; sid:1000003; rev:1;)

# Alert on any UDP traffic to port 53 (DNS)
alert udp any any -> any 53 (msg:"UDP Traffic to DNS port 53 detected"; sid:1000004; rev:1;)

# Alert on any TCP traffic to port 22 (SSH)
alert tcp any any -> any 22 (msg:"TCP Traffic to SSH port 22 detected"; sid:1000005; rev:1;)

5. Test Snort Configuration

Test the configuration to ensure there are no syntax errors:

sudo snort -T -c /etc/snort/snort.conf

6. Run Snort

Run Snort in IDS mode:

To monitor multiple network interfaces with Snort, You can run Snort multiple times, each time specifying a different interface:

snort -A console -c /etc/snort/snort.conf -i <network-interface>
snort -A console -c /etc/snort/snort.conf -i ens33 &
snort -A console -c /etc/snort/snort.conf -i eth1 &

Replace <network-interface> with your network interface, for example, ens33.

  • Sample output of result

7. Automate Snort Startup

To ensure Snort starts on boot, create a systemd service file.

Create Systemd Service File:

sudo nano /etc/systemd/system/snort33.service

Add the following content:

[Unit]
Description=Snort NIDS
After=network.target

[Service]
ExecStart=/usr/sbin/snort -c /etc/snort/snort.conf -i ens33
ExecReload=/bin/kill -HUP $MAINPID
Restart=always
RestartSec=5
[Install]
WantedBy=multi-user.target

Create another Systemd Service File for vpn tunnel [connecting branch and main servers]:

sudo nano /etc/systemd/system/tun.service

Add the following content:

[Unit]
Description=Snort tunnel NIDS
After=network.target

[Service]
ExecStart=/usr/sbin/snort -c /etc/snort/snort.conf -i tun0
ExecReload=/bin/kill -HUP $MAINPID
Restart=always
RestartSec=5
[Install]
WantedBy=multi-user.target

Start the services

systemctl daemon-reload
systemctl enable tun.service
systemctl start tun.service
systemctl enable snort33.service
systemctl start snort33.service
systemctl status snort33.service

8. Monitor Snort Logs

Snort logs its alerts to /var/log/snort/snort.alert.fast by default. You can monitor this file for suspicious activity.

tail -f /var/log/snort/snort.alert.fast

By following these steps, Snort will monitor your network traffic for suspicious activity and log alerts based on the rules defined. You can optionally adjust and expand the rules and configuration to match the specific requirements and threats relevant to your environment.

Configure NMAP:

To use nmap for network scanning and to ensure that all expected services are running and accessible, follow these steps:

1. Install Nmap on branch server

First, make sure nmap is installed on your system. You can install it using the package manager for your distribution:

For Debian/Ubuntu-based systems:

sudo apt-get update
sudo apt-get install nmap

2. Basic Network Scan

To scan an entire network for live hosts and open ports, use:

nmap -sP 192.168.5.0/24

or

nmap -sn 192.168.5.0/24

This will perform a “ping scan” to determine which hosts are online.

3. Service Detection

To detect services running on specific hosts or an entire network, use the following command:

nmap -sV 192.168.5.0/24

This performs a service/version detection scan to determine what services and versions are running on open ports.

4. Port Scanning

To scan for open ports on a specific host:

nmap -p- 192.168.5.10

This scans all 65535 ports. You can specify a range of ports:

Command below scans for port 22 and port 80 on the server

nmap -p 22,80,443 192.168.5.15

5. Aggressive Scan

An aggressive scan performs a detailed scan including service detection, OS detection, and more:

nmap -A 192.168.5.15

This will give you extensive information about the host, including open ports, services, OS, and more.

7. Scan Multiple Hosts

To scan multiple hosts or subnets, list them in the command:

sudo nmap -p 80,22 192.168.5.10 192.168.5.15 192.168.1.10

8. Output Formats

For easier analysis, you can save the results in various formats:

  • Normal Output:
sudo nmap -oN scan_results.txt 192.168.5.15 192.168.5.10

9. Schedule Regular Scans

To ensure services are consistently monitored, you can set up a cron job to run nmap regularly.

Edit the crontab with:

crontab -e

Add an entry to run nmap at a regular interval, e.g., daily at midnight:

0 0 * * * /usr/bin/nmap -sP 192.168.5.0/24 > /var/log/nmap_daily_scan.log

These steps will help ensure all expected services are running and accessible on your network.

Capture network packets withtcpdump

Capturing and analyzing network traffic using tcpdump and Wireshark can help you verify that your VPN is working correctly and identify any potential issues. Below are the steps to perform this task on both the main and branch servers.

Installtcpdump if not already pre-installed.

For Debian/Ubuntu-based systems:

sudo apt-get update
sudo apt-get install tcpdump

Capture Traffic

You need to capture traffic on the interfaces involved in the VPN connection. For example, my VPN setup interface is tun0, I can use the following command:

sudo tcpdump -i tun0 -w vpn_traffic.pcap

  • -i tun0: Specifies the interface to capture traffic on (replace tun0 with your own actual VPN interface).

  • -w vpn_traffic.pcap: Writes the captured packets to a file named vpn_traffic.pcap.

You can also filter the traffic by IP address or port if you want to narrow down the capture:

sudo tcpdump -i tun0 host 10.0.0.1 -w vpn_traffic.pcap

Capture on Both Servers

Run the above tcpdump commands on both the main and branch servers to capture the relevant traffic.

I will also run a ping and ssh commands across both servers in another session so that ICMP and SSH packets can be captured in the tcpdump process.

ping 10.0.0.2 # From the main server 
ping 10.0.0.1 # From the branch server 
ssh user@10.0.0.2 # From the main server 
ssh user@10.0.0.1 # From the branch server

Step 2: Transfer the Capture Files

After capturing the traffic, transfer the .pcap files to your local machine for analysis. You can use scp (secure copy) to do this:

scp user@branch_server:/path/to/vpn_traffic.pcap /local/path/
scp user@main_server:/path/to/vpn_traffic.pcap /local/path/

Step 3: Analyze Traffic with Wireshark

Install Wireshark

Wireshark is available for Windows, macOS, and Linux. You can download it from the official Wireshark website.

Open captured files in Wireshark and analyze:

  1. Open Wireshark.

  2. Click File -> Open and select the transferred .pcap files to open.

  3. Use Wireshark’s display filters to focus on specific traffic. For example, to display only traffic between two IP addresses, enter below strings in the search bar display fileter, and click on the blue arrow to the right to apply.

  • ip.addr == 10.0.0.1 && ip.addr == 10.0.0.2

4. Inspect VPN Traffic: Look at the traffic on the tun0 interface (or your VPN interface) to ensure packets are being transmitted and received correctly.

5. Check for Anomalies: Look for retransmissions, malformed packets, or any other unusual activity.

By following these steps, you can capture and analyze the network traffic on both your main and branch servers, verify the VPN functionality, and troubleshoot any potential issues using tcpdump and Wireshark.

AUTOMATION

Configure Ansible Playbooks

I will also install ansible and Based on the manual steps I have used to initially setup the configuration, I will create Ansible playbooks to automate the setup for DHCP, DNS, NTP, and basic firewall rules.

Here is a code file containing the steps to install and configure Ansible on an Ubuntu system for this environment.

#!/bin/bash

# Update package index
echo "Updating package index..."
sudo apt update

# Install Ansible
echo "Installing Ansible..."
sudo apt install ansible -y

# Optionally, install the latest version of Ansible via PPA
echo "Adding Ansible PPA..."
sudo add-apt-repository ppa:ansible/ansible -y
sudo apt update
sudo apt install ansible -y

# Verify Ansible installation
echo "Verifying Ansible installation..."
ansible --version

# Create and configure the inventory file
echo "Configuring Ansible inventory file..."
sudo tee /etc/ansible/hosts > /dev/null <<EOL
[main_server]
192.168.1.10

[branch_server]
192.168.5.10

[client_server]
192.168.5.15
EOL

# Edit the Ansible configuration file
echo "Editing Ansible configuration file..."
sudo tee /etc/ansible/ansible.cfg > /dev/null <<EOL
[defaults]
inventory = /etc/ansible/hosts
remote_user = your_user
private_key_file = /path/to/private/key
EOL

# Test Ansible configuration
echo "Testing Ansible configuration..."
ansible all -m ping

echo "Ansible installation and configuration complete!"

Instructions for Using the Script

Save the Script:

Save the above script to a file, e.g., install_ansible.sh.

Make the Script Executable:

chmod +x install_ansible.sh

Run the Script:

./install_ansible.sh

This script handles the installation of Ansible, adds the Ansible PPA if desired, sets up the inventory file, and configures the Ansible configuration file. Adjust the remote_user and private_key_file in the Ansible configuration file according to your setup.

Below are the Ansible playbooks for each service:

1. DHCP Configuration

Playbook:dhcp.yml

---
- name: Configure DHCP Server
  hosts: branch_server
  become: yes
  tasks:
    - name: Install DHCP server
      apt:
        name: isc-dhcp-server
        state: present
        update_cache: yes

    - name: Configure DHCP server
      copy:
        dest: /etc/dhcp/dhcpd.conf
        content: |
          option domain-name "company.local";
          default-lease-time 600;
          max-lease-time 7200;

          subnet 192.168.5.0 netmask 255.255.255.0 {
              range 192.168.5.50 192.168.5.100;
              option routers 192.168.5.10;
              option domain-name-servers 192.168.5.10, 8.8.8.8;
          }

    - name: Specify network interface for DHCP
      lineinfile:
        path: /etc/default/isc-dhcp-server
        regexp: '^INTERFACESv4='
        line: 'INTERFACESv4="ens33"'

    - name: Restart DHCP server
      service:
        name: isc-dhcp-server
        state: restarted
        enabled: yes

2. DNS Configuration

Playbook:dns.yml

---
- name: Configure DNS Server
  hosts: main_server
  become: yes
  tasks:
    - name: Install BIND9
      apt:
        name: "{{ item }}"
        state: present
        update_cache: yes
      with_items:
        - bind9
        - bind9utils
        - bind9-doc

    - name: Configure BIND9 named.conf.local
      copy:
        dest: /etc/bind/named.conf.local
        content: |
          zone "abc.local" {
              type master;
              file "/etc/bind/db.abc.local";
          };

          zone "5.168.192.in-addr.arpa" {
              type master;
              file "/etc/bind/db.192.168.5";
          };

    - name: Create forward zone file for abc.local
      copy:
        dest: /etc/bind/db.abc.local
        content: |
          $TTL    604800
          @       IN      SOA     main.abc.local. admin.abc.local. (
                                   2         ; Serial
                              604800         ; Refresh
                               86400         ; Retry
                             2419200         ; Expire
                              604800 )       ; Negative Cache TTL
          ;
          @       IN      NS      main.abc.local.
          main    IN      A       192.168.1.10
          branch  IN      A       192.168.5.10
          client  IN      A       192.168.5.15

    - name: Create reverse zone file for 192.168.5.x
      copy:
        dest: /etc/bind/db.192.168.5
        content: |
          $TTL    604800
          @       IN      SOA     main.abc.local. admin.abc.local. (
                                   2         ; Serial
                              604800         ; Refresh
                               86400         ; Retry
                             2419200         ; Expire
                              604800 )       ; Negative Cache TTL
          ;
          @       IN      NS      main.abc.local.
          10      IN      PTR     branch.abc.local.
          15      IN      PTR     client.abc.local.

    - name: Configure BIND9 forwarders
      lineinfile:
        path: /etc/bind/named.conf.options
        insertafter: '{'
        line: |
          forwarders {
              8.8.8.8;  // Google's DNS
              8.8.4.4;  // Google's DNS
          };

    - name: Restart BIND9
      service:
        name: bind9
        state: restarted
        enabled: yes

- name: Configure DNS Server
  hosts: branch_server
  become: yes
  tasks:
    - name: Install BIND9
      apt:
        name: "{{ item }}"
        state: present
        update_cache: yes
      with_items:
        - bind9
        - bind9utils
        - bind9-doc

    - name: Configure BIND9 named.conf.local as slave
      copy:
        dest: /etc/bind/named.conf.local
        content: |
          zone "abc.local" {
              type slave;
              file "/var/cache/bind/db.abc.local";
              masters { 192.168.1.10; };
          };

          zone "5.168.192.in-addr.arpa" {
              type slave;
              file "/var/cache/bind/db.192.168.5";
              masters { 192.168.1.10; };
          };

    - name: Configure BIND9 forwarders
      lineinfile:
        path: /etc/bind/named.conf.options
        insertafter: '{'
        line: |
          forwarders {
              192.168.1.10;  // Main server's IP
          };

    - name: Restart BIND9
      service:
        name: bind9
        state: restarted
        enabled: yes

3. NTP Configuration (using chrony) with iptables rules

Playbook:ntp.yml

---
- name: Configure NTP with Chrony
  hosts: branch_server
  become: yes
  tasks:
    - name: Install chrony
      apt:
        name: chrony
        state: present
        update_cache: yes

    - name: Configure chrony to allow network
      lineinfile:
        path: /etc/chrony/chrony.conf
        line: 'allow 192.168.5.0/24'
        state: present

    - name: Restart chrony
      service:
        name: chrony
        state: restarted
        enabled: yes

    - name: Add iptables rules for chrony
      iptables:
        chain: INPUT
        protocol: udp
        destination_port: 123
        jump: ACCEPT

    - name: Add iptables rules for chrony output
      iptables:
        chain: OUTPUT
        protocol: udp
        source_port: 123
        jump: ACCEPT

    - name: Save iptables rules
      command: iptables-save > /etc/iptables/rules.v4

- name: Configure NTP with Chrony on Client Server
  hosts: client_server
  become: yes
  tasks:
    - name: Install chrony
      apt:
        name: chrony
        state: present
        update_cache: yes

    - name: Configure chrony to use branch server
      lineinfile:
        path: /etc/chrony/chrony.conf
        line: 'server 192.168.5.10 iburst'
        state: present

    - name: Restart chrony
      service:
        name: chrony
        state: restarted
        enabled: yes

    - name: Add iptables rules for chrony
      iptables:
        chain: INPUT
        protocol: udp
        destination_port: 123
        jump: ACCEPT

    - name: Add iptables rules for chrony output
      iptables:
        chain: OUTPUT
        protocol: udp
        source_port: 123
        jump: ACCEPT

    - name: Save iptables rules
      command: iptables-save > /etc/iptables/rules.v4

4. Basic Firewall Rules with additional iptables rules

Playbook:firewall.yml

---
- name: Configure basic firewall rules and iptables
  hosts: branch_server
  become: yes
  tasks:
    - name: Install UFW
      apt:
        name: ufw
        state: present
        update_cache: yes

    - name: Allow SSH
      ufw:
        rule: allow
        name: 'OpenSSH'

    - name: Allow DHCP
      ufw:
        rule: allow
        port: 67
        proto: udp

    - name: Allow DNS
      ufw:
        rule: allow
        port: 53
        proto: udp

    - name: Allow NTP
      ufw:
        rule: allow
        port: 123
        proto: udp

    - name: Enable UFW
      ufw:
        state: enabled

    - name: Add NAT iptables rule
      iptables:
        chain: POSTROUTING
        table: nat
        out_interface: ens38
        jump: MASQUERADE

    - name: Add forward rule for related/established connections
      iptables:
        chain: FORWARD
        in_interface: ens38
        out_interface: ens33
        match: state
        state: RELATED,ESTABLISHED
        jump: ACCEPT

    - name: Add forward rule to allow traffic from ens33 to ens38
      iptables:
        chain: FORWARD
        in_interface: ens33
        out_interface: ens38
        jump: ACCEPT

    - name: Save iptables rules
      command: iptables-save > /etc/iptables/rules.v4
Inventory host File
[main_server]
192.168.1.10[branch_server]
192.168.5.10[client_server]
192.168.5.15[all]
192.168.1.10
192.168.5.10
192.168.5.15

Run the Playbooks

To execute the playbooks, use the following commands:

ansible-playbook -i hosts dhcp.yml
ansible-playbook -i hosts dns.yml
ansible-playbook -i hosts ntp.yml
ansible-playbook -i hosts firewall.yml

These playbooks will automate the configuration of DHCP, DNS, NTP with iptables rules, and basic firewall rules on the specified servers.

Infrastructure as code:

To use Vagrant to provision a VM on VMware Workstation, simulating an additional branch server and client system, follow these steps. This guide will cover installing the necessary tools, setting up Vagrant, and creating Vagrantfiles to provision the VMs.

Prerequisites

  1. VMware Workstation: Ensure VMware Workstation is installed.

  2. Vagrant: Install Vagrant on your system. https://developer.hashicorp.com/vagrant/install

  3. Vagrant VMware Utility: Install the Vagrant VMware Utility. https://developer.hashicorp.com/vagrant/docs/providers/vmware/vagrant-vmware-utility

  4. Vagrant VMware Desktop Plugin: Install the Vagrant VMware Desktop plugin.

  5. Run the following command in your terminal :

vagrant plugin install vagrant-vmware-desktop

Steps to Provision VMs with Vagrant

1. Create a Directory for Your Vagrant Project

Create a directory for your Vagrant project, and navigate into it:

mkdir vagrant
cd vagrant

2. Initialize Vagrant

Initialize Vagrant in the directory:

vagrant init

This will create a Vagrantfile in the directory. You will modify the Vagrantfiles to contain the resource creation for the 3 VMs. Check for latest OS version at https://app.vagrantup.com/generic

To add additional configurations to each of the VMs, you can use a shell provisioner in the Vagrantfile. Below, I’ve created a script to set up DHCP, iptables, DNS, Chrony, and firewall rules. The script will be called provision.sh, and it will be referenced in the Vagrantfile for each VM.

Vagrant.configure("2") do |config|
  # Define the "main" VM
  config.vm.define "main" do |main|
    main.vm.box = "generic/ubuntu2310"
    main.vm.network "private_network", ip: "192.168.1.10"
    main.vm.hostname = "main"

    main.vm.provider "vmware_desktop" do |v|
      v.vmx["memsize"] = "2048"
      v.vmx["numvcpus"] = "2"
      v.vmx["disk.size"] = "20000"
    end

    main.vm.provision "shell", path: "C:/Users/balog/Desktop/vagrant-branch-client/provision.sh"
  end

  # Define the "branch" VM
  config.vm.define "branch" do |branch|
    branch.vm.box = "generic/ubuntu2310"
    branch.vm.network "private_network", ip: "192.168.5.10"
    branch.vm.hostname = "branch"

    branch.vm.provider "vmware_desktop" do |v|
      v.vmx["memsize"] = "2048"
      v.vmx["numvcpus"] = "2"
      v.vmx["disk.size"] = "20000"
    end

    branch.vm.provision "shell", path: "C:/Users/balog/Desktop/vagrant-branch-client/provision.sh"
  end

  # Define the "client" VM
  config.vm.define "client" do |client|
    client.vm.box = "generic/ubuntu2310"
    client.vm.network "private_network", type: "dhcp"
    client.vm.hostname = "client"

    client.vm.provider "vmware_desktop" do |v|
      v.vmx["memsize"] = "2048"
      v.vmx["numvcpus"] = "2"
      v.vmx["disk.size"] = "20000"
    end

    client.vm.provision "shell", path: "C:/Users/balog/Desktop/vagrant-branch-client/provision.sh"
  end
end

Script provision for vagrant file

#!/bin/bash

# Update and install necessary packages
sudo apt-get update

# Install packages based on hostname
if [[ $(hostname) == "main" || $(hostname) == "branch" ]]; then
    sudo apt-get install -y bind9 bind9utils bind9-doc tinc

    if [[ $(hostname) == "branch" ]]; then
        sudo apt-get install -y isc-dhcp-server chrony
    fi
elif [[ $(hostname) == "client" ]]; then
    sudo apt-get install -y tinc chrony
fi

# Configure BIND9 for Main Server
if [[ $(hostname) == "main" ]]; then
    # BIND9 Configuration
    cat <<EOF | sudo tee /etc/bind/named.conf.local
zone "abc.local" {
    type master;
    file "/etc/bind/db.abc.local";
};

zone "5.168.192.in-addr.arpa" {
    type master;
    file "/etc/bind/db.192.168.5";
};
EOF

    cat <<EOF | sudo tee /etc/bind/db.abc.local
\$TTL    604800
@       IN      SOA     main.abc.local. admin.abc.local. (
                           2         ; Serial
                      604800         ; Refresh
                       86400         ; Retry
                     2419200         ; Expire
                      604800 )       ; Negative Cache TTL
;
@       IN      NS      main.abc.local.
main    IN      A       192.168.1.10
branch  IN      A       192.168.5.10
client  IN      A       192.168.5.15
EOF

    cat <<EOF | sudo tee /etc/bind/db.192.168.5
\$TTL    604800
@       IN      SOA     main.abc.local. admin.abc.local. (
                           2         ; Serial
                      604800         ; Refresh
                       86400         ; Retry
                     2419200         ; Expire
                      604800 )       ; Negative Cache TTL
;
@       IN      NS      main.abc.local.
10      IN      PTR     branch.abc.local.
15      IN      PTR     client.abc.local.
EOF

    cat <<EOF | sudo tee /etc/bind/named.conf.options
options {
    directory "/var/cache/bind";

    forwarders {
        8.8.8.8;
        8.8.4.4;
    };

    dnssec-validation auto;

    listen-on-v6 { any; };
};
EOF

    sudo systemctl start bind9

# Configure BIND9 for Branch Server
elif [[ $(hostname) == "branch" ]]; then
    # BIND9 Configuration
    cat <<EOF | sudo tee /etc/bind/named.conf.local
zone "abc.local" {
    type slave;
    file "/var/cache/bind/db.abc.local";
    masters { 192.168.1.10; };
};

zone "5.168.192.in-addr.arpa" {
    type slave;
    file "/var/cache/bind/db.192.168.5";
    masters { 192.168.1.10; };
};
EOF

    cat <<EOF | sudo tee /etc/bind/named.conf.options
options {
    directory "/var/cache/bind";

    forwarders {
        192.168.1.10;
    };

    dnssec-validation auto;

    listen-on-v6 { any; };
};
EOF

    sudo systemctl start bind9

    # Configure DHCP Server
    cat <<EOF | sudo tee /etc/dhcp/dhcpd.conf
option domain-name "company.local";
default-lease-time 600;
max-lease-time 7200;

subnet 192.168.5.0 netmask 255.255.255.0 {
    range 192.168.5.15 192.168.5.50;
    option routers 192.168.5.10;
    option domain-name-servers 192.168.5.10, 8.8.8.8;
}
EOF

    sudo sed -i 's/INTERFACESv4=""/INTERFACESv4="ens33"/' /etc/default/isc-dhcp-server

    sudo systemctl start isc-dhcp-server

    # Enable IP Forwarding and configure NAT
    sudo sed -i 's/#net.ipv4.ip_forward=1/net.ipv4.ip_forward=1/' /etc/sysctl.conf
    sudo sysctl -p
    sudo iptables -t nat -A POSTROUTING -o ens38 -j MASQUERADE
    sudo iptables -A FORWARD -m state --state RELATED,ESTABLISHED -j ACCEPT
    sudo iptables -A FORWARD -j ACCEPT
    sudo sh -c "iptables-save > /etc/iptables/rules.v4"
#    sudo apt-get install iptables-persistent -y

    # Configure Tinc VPN for Branch Server
    sudo mkdir -p /etc/tinc/vpn/hosts

    cat <<EOF | sudo tee /etc/tinc/vpn/tinc.conf
Name = branch
AddressFamily = ipv4
Interface = tun0
ConnectTo = main
EOF

    cat <<EOF | sudo tee /etc/tinc/vpn/tinc-up
#!/bin/sh
ifconfig \$INTERFACE 10.0.0.2 netmask 255.255.255.0
EOF

    cat <<EOF | sudo tee /etc/tinc/vpn/tinc-down
#!/bin/sh
ifconfig \$INTERFACE down
EOF

    sudo chmod +x /etc/tinc/vpn/tinc-up /etc/tinc/vpn/tinc-down

    cat <<EOF | sudo tee /etc/tinc/vpn/hosts/branch
Address = 192.168.5.10
Subnet = 10.0.0.2/32
EOF

    sudo tincd -n vpn -K4096
    sudo systemctl enable tinc@vpn
    sudo systemctl start tinc@vpn

# Client Configuration
elif [[ $(hostname) == "client" ]]; then
    # Configure Netplan
    cat <<EOF | sudo tee /etc/netplan/01-netcfg.yaml
network:
  version: 2
  ethernets:
    ens33:
      dhcp4: yes
      routes:
        - to: default
          via: 192.168.5.10
      nameservers:
        addresses:
          - 192.168.5.10
          - 8.8.8.8
EOF

    sudo netplan apply

    # Configure Chrony
    sudo sed -i '/pool /d' /etc/chrony/chrony.conf
    echo "server branch.abc.local iburst" | sudo tee -a /etc/chrony/chrony.conf
    sudo systemctl start chrony
    sudo systemctl enable chrony
fi

# Configure Chrony on Branch Server
if [[ $(hostname) == "branch" ]]; then
    sudo sed -i '/pool /d' /etc/chrony/chrony.conf
    echo "server main.abc.local prefer iburst" | sudo tee -a /etc/chrony/chrony.conf
    sudo systemctl start chrony
    sudo systemctl enable chrony
fi

# Common Configuration
# Set hostname and update /etc/hosts
sudo hostnamectl set-hostname $(hostname).abc.local
cat <<EOF | sudo tee -a /etc/hosts
192.168.1.10 main.abc.local
192.168.5.10 branch.abc.local
192.168.5.15 client.abc.local
EOF

This provisioner will set up BIND9 for both the main and branch servers, configure Tinc VPN, set up DHCP on the branch server, and provide internet access to the client.

5. Start the VMs

vagrant up

6. Verify the VMs

vagrant status

Use vmware workstation to discover and manage the running VMs

  • Open vmware workstation, click on File, then select option scan for virtual machines

  • Browse for the installation location of the created VMs, follow the instructions on screen and launch it.

Verify the VMs on the VMware workstation

Other vagrant commands

vagrant halt #to halt the VM
vagrant destroy #to remove the VM setup
vagrant validate #to validate the configuration files in the vagrant directory
vagrant status #show status of vagrant VMs

In Conclusion

Completing this scenario will provide you with practical experience in network administration, enabling you to apply the concepts and commands learned to address real-world challenges faced by Cloud Engineers and System Administrators.

0
Subscribe to my newsletter

Read articles from Seun B directly inside your inbox. Subscribe to the newsletter, and don't miss out.

Written by

Seun B
Seun B

Automation Enthusiast