Skip to content

PiHole HA Setup⚑

What are we trying to achieve?⚑

We want to have two PiHole instances that share the same ip address. If one of the instances fails the other one will take over the ip address.

They will also share the same gravity database so you only have to update the gravity database on one of the instances.

Why do we want this?⚑

If you have a PiHole instance running on a Raspberry Pi and it fails you will have to wait until you can fix it. This means manually changing the dns server on all your devices or trying to change the dhcp server to point to a different dns server. With this setup you will have a backup PiHole instance that will take over the ip address of the primary instance when it fails.

Requirements⚑

  • Two PiHole instances
  • A ip address that is not used by any other device on your network
  • Both PiHole instances must be on the same network
  • Both PiHole instances must have the same version, check with pihole -v, update with pihole -up
  • They may need to be reconfigured with pihole -r to get them to work properly

Setup keepalive⚑

Install required packages⚑

sudo apt install keepalived

Configure keepalived⚑

Script to check if local instance is running

cat << EOF > /usr/local/bin/check-local-pihole
#!/bin/sh

RUNNING=$(ps -aux | grep pihole-FTL | grep -v grep)
exit $?
EOF

chmod +x /usr/local/bin/check-local-pihole

Configure keepalived on the primary instance⚑

cat << EOF > /etc/keepalived/keepalived.conf
vrrp_script chk_local_pihole {
    script "/usr/local/bin/check-local-pihole" # (1)!
    interval 5
    weight -100
}

global_defs {
  router_id pihole-01 # (2)!
  script_user root
  enable_script_security
}

vrrp_instance PIHOLE {
  state MASTER # (3)!
  interface eth0 # (4)!
  virtual_router_id 20 # (5)!
  priority 150
  advert_int 1
  unicast_src_ip 192.168.3.21 # (6)!
  unicast_peer {
    192.168.3.22 # (7)!
  }

  authentication {
    auth_type PASS
    auth_pass piholedns # (8)!
  }

  virtual_ipaddress {
    192.168.3.20/23 # (9)!
  }

  track_script {
    chk_local_pihole
  }
}
EOF
  1. Path to the script that checks if the local instance is running
  2. Name of the router
  3. State of the instance, MASTER or BACKUP
  4. Interface to use
  5. virtual_router_id, must be the same on both instances
  6. unicast_src_ip, ip address of the local instance
  7. unicast_peer, ip address of the remote instance
  8. auth_pass, must be the same on both instances
  9. virtual_ipaddress, ip address that will be used by the clients and shared between the two instances
Configure keepalived on the secondary instance⚑
cat << EOF > /etc/keepalived/keepalived.conf
vrrp_script chk_pihole {
  script "/usr/local/bin/check-local-pihole"
  interval 1
  weight -100
}

global_defs {
  router_id pihole-02
  script_user root
  enable_script_security
}

vrrp_instance PIHOLE {
  state BACKUP
  interface eth0
  virtual_router_id 20
  priority 140
  advert_int 1
  unicast_src_ip 192.168.3.22
  unicast_peer {
    192.168.3.21
  }

  authentication {
    auth_type PASS
    auth_pass piholedns
  }

  virtual_ipaddress {
    192.168.3.20/23
  }

  track_script {
    chk_local_pihole
  }
}
EOF

Start keepalived⚑

Run on both instances

systemctl enable --now keepalived.service

Test keepalived⚑

  1. Check if the pihole portal is reachable through the virtual ip address
  2. Check if the virtual ip address is assigned to the primary instance by watching the output of ip a on both instances or looking at the pihole dashboard in the top right corner
  3. Restart the pihole service on the primary instance and check if the virtual ip address is assigned to the secondary instance
  4. Check if the primary instance will take over the virtual ip address when it is restarted

Sync gravity database⚑

Install required packages⚑

apt update && apt install sqlite3 sudo git cron rsync ssh

Install gravity sync script⚑

We will use gravity-sync to sync the gravity database between the two instances.

Install gravity-sync on both instances and follow the instructions.

curl -sSL https://raw.githubusercontent.com/vmstan/gs-install/main/gs-install.sh | bash

You can always reset the configuration with gravity-sync config

Push gravity database to secondary instance⚑

Run the following command on the primary instance to push the gravity database to the secondary instance.

gravity-sync push

Automate gravity database sync⚑

Run the following command on both instances to create a systemd timer that will sync the gravity database every 5 minutes.

gravity-sync automate

You can check the status of the timer with systemctl status gravity-sync.timer.
And you can check the logs with journalctl -u gravity-sync.service.

With gravity-sync automate hour the timer will sync the gravity database every hour.