x386.org

Keepalived

High-availability clustering

0. Specs

0.0. The What

Keepalived is a Linux daemon that provides high availability and load balancing using the Virtual Router Redundancy Protocol (VRRP). It is commonly used to ensure service continuity by automatically failing over to a backup server if the primary server becomes unavailable.

In this tutorial, we'll configure two servers with various Keepalived scenarios to demonstrate different high-availability configurations.

0.1. The Environment

  • Server OS: Debian 13 or Ubuntu 26.04 Server
  • Virtual IP Address (VIP): 192.168.1.240

0.2. Scenarios

  • Scenario 1 - Simple Mode: Two servers where the first server holds the VIP. If it fails, the second server takes over the VIP. When the first server recovers, it reclaims the VIP (preemptive mode).

  • Scenario 2 - Nonpreemptive Mode: Similar to Scenario 1, but when the first server recovers, it does not automatically reclaim the VIP.

  • Scenario 3 - Apache High Availability: Both servers run Apache. The VIP fails over to the backup server only if Apache becomes unreachable on the primary server, not just if the server itself is down.

  • Scenario 4 - Active/Active VIPs: Two servers and two VIPs (192.168.1.240 and 192.168.1.241). Each server primarily holds one VIP, but can take over the other VIP if the peer server fails.

0.3. Sources


1. Scenario 1 - Simple Mode

1.1. Install keepalived

Install on both servers (primary and backup):

sudo apt update
sudo apt -y install keepalived

Note: After installation, the service may attempt to start but will fail until a valid configuration is provided.

1.2. First Server Configuration

The configuration directory is /etc/keepalived/. Create the main configuration file:

sudo nano /etc/keepalived/keepalived.conf

Add the following configuration:

global_defs {
    router_id main   # Unique identifier for this node in logs
}

vrrp_instance VI_1 {
    state MASTER          # This node starts as the primary
    interface enp0s3      # Network interface name - CHANGE TO YOURS
    virtual_router_id 51  # Must be identical across all nodes in the group
    priority 100          # Election priority (higher = more preferred)
    advert_int 1          # Advertisement interval in seconds

    virtual_ipaddress {
        192.168.1.240/24  # Virtual IP to manage (include netmask)
    }
}

1.3. Second Server Config

Similar but with a few differences.

Create the configuration file on the backup server:

sudo nano /etc/keepalived/keepalived.conf

Fill as below:

global_defs {
    router_id backup    # Unique identifier for this node in logs
}

vrrp_instance VI_1 {
    state BACKUP          # This node starts as backup
    interface enp0s3      # Network interface name - CHANGE TO YOURS
    virtual_router_id 51  # Must match the primary server
    priority 90           # Lower priority than primary
    advert_int 1

    virtual_ipaddress {
        192.168.1.240/24
    }
}

1.5. Start keepalived

Run on both servers

sudo systemctl start keepalived

You can check the status of your cluster

systemctl status -l keepalived

Testing: You can test failover by stopping the primary server (sudo systemctl stop keepalived or shutting it down) and observing the VIP migration to the backup server.

2. Scenario 2 - Nonpreemptive Mode

2.0. Overview

Keepalived defaults to preemptive mode, where a higher-priority server that recovers will automatically reclaim the VIP. In nonpreemptive mode, once a backup server takes over the VIP, it will keep it even if the original primary server recovers.

To implement this, we add the nopreempt directive and remove the state declarations from both configurations.

2.1. Update Configurations

On the primary server, edit the configuration:

sudo nano /etc/keepalived/keepalived.conf

Update with the following content:

global_defs {
    router_id main
}

vrrp_instance VI_1 {
    interface enp0s3
    virtual_router_id 51
    priority 100
    advert_int 1
    nopreempt               # Disable automatic failback

    virtual_ipaddress {
        192.168.1.240/24 
    }
}

On the backup server, edit the configuration:

sudo nano /etc/keepalived/keepalived.conf

Update with the following content:

global_defs {
    router_id backup
}

vrrp_instance VI_1 {
    interface enp0s3
    virtual_router_id 51
    priority 90
    advert_int 1
    nopreempt           # Disable automatic failback

    virtual_ipaddress {
        192.168.1.240/24
    }
}

2.2. Restart Services

Restart Keepalived on both servers to apply the changes:

sudo systemctl restart keepalived

3. Scenario 3 - Apache High Availability

3.0. Overview

This scenario provides application-level high availability for Apache web server. The VIP will fail over only if Apache service itself becomes unavailable, not just if the server is unreachable.

We'll create a custom health check script and integrate it with Keepalived.

3.1. Create Health Check Script

On both servers, create the check script:

sudo nano /etc/keepalived/check_apache.sh

Add the following content:

#!/bin/bash
# Simple check: is Apache running?
# Works for both Apache2 and httpd-based systems.

if pidof apache2 >/dev/null 2>&1; then
    exit 0
else
    exit 1
fi

Make the script executable:

sudo chmod +x /etc/keepalived/check_apache.sh

3.2. Install and Verify Apache

Ensure Apache is installed and running on both servers:

sudo apt update
sudo apt install -y apache2

3.3. Update Keepalived Configuration

On the primary server, update the configuration:

sudo nano /etc/keepalived/keepalived.conf

Use the following configuration:

global_defs {
    router_id main
    script_user root           # User context for script execution
    enable_script_security     # Enable script security features
}

vrrp_script chk_apache {
    script "/etc/keepalived/check_apache.sh"
    interval 2        # Check every 2 seconds
    timeout 2         # Script must complete within 2 seconds
    fall 2            # 2 consecutive failures trigger a fault
    rise 2            # 2 consecutive successes clear a fault
    weight -20        # Deduct 20 from priority if Apache fails
}

vrrp_instance VI_1 {
    interface enp0s3
    virtual_router_id 51
    priority 100
    advert_int 1

    virtual_ipaddress {
        192.168.1.240/24
    }

    track_script {
        chk_apache    # Monitor Apache service health
    }
}

On the backup server, update the configuration:

sudo nano /etc/keepalived/keepalived.conf

Use the following configuration:

global_defs {
    router_id backup
    script_user root
    enable_script_security
}

vrrp_script chk_apache {
    script "/etc/keepalived/check_apache.sh"
    interval 2
    timeout 2
    fall 2
    rise 2
    weight -20
}

vrrp_instance VI_1 {
    interface enp0s3
    virtual_router_id 51
    priority 90
    advert_int 1

    virtual_ipaddress {
        192.168.1.240/24
    }

    track_script {
        chk_apache
    }
}

3.4. Restart Services

Restart Keepalived on both servers:

sudo systemctl restart keepalived

Testing: Stop Apache on the primary server (sudo systemctl stop apache2) and observe the VIP failing over to the backup server.

Note: For Nginx or other services, modify the check script accordingly (e.g., check for nginx instead of apache2).

4. Scenario 4 - Active/Active VIPs

4.0. Overview

This configuration provides active-active load distribution with two VIPs: - VIP1: 192.168.1.240 (primarily on Server 1) - VIP2: 192.168.1.241 (primarily on Server 2)

Both servers remain active simultaneously, but each can take over the other's VIP if a failure occurs.

4.1. Update Keepalived Configuration

On the first server, update the configuration:

sudo nano /etc/keepalived/keepalived.conf

On the first server, change as below:

global_defs {
    router_id first
}

vrrp_instance VIP_1 {
    state MASTER
    interface enp0s3
    virtual_router_id 100    # Unique ID for first VIP group
    priority 100
    advert_int 1
    virtual_ipaddress {
        192.168.1.240/24    # Primary VIP for this server
    }
}

vrrp_instance VIP_2 {
    state BACKUP
    interface enp0s3
    virtual_router_id 101    # Unique ID for second VIP group
    priority 90
    advert_int 1
    virtual_ipaddress {
        192.168.1.241/24    # Secondary VIP (backup role)
    }
}

On the second server, update the configuration:

sudo nano /etc/keepalived/keepalived.conf

Use the following configuration:

global_defs {
    router_id second
}

vrrp_instance VIP_1 {
    state BACKUP
    interface enp0s3
    virtual_router_id 100
    priority 90
    advert_int 1
    virtual_ipaddress {
        192.168.1.240/24    # Secondary VIP (backup role)
    }
}

vrrp_instance VIP_2 {
    state MASTER
    interface enp0s3
    virtual_router_id 101
    priority 100
    advert_int 1
    virtual_ipaddress {
        192.168.1.241/24    # Primary VIP for this server
    }
}

4.2. Restart Services

Restart Keepalived on both servers:

sudo systemctl restart keepalived

4.3. Optional: Add Service Health Checks

You can enhance this configuration by adding health check scripts (as shown in Scenario 3) to each vrrp_instance block to trigger failover based on application health rather than just server availability.

x386.org