network docker0 interface broke role; multiple interfaces

[A-SOM]

Bug description

The presence of multiple interfaces, such as Docker, can break the operation of playbook 2.3.2.
example: pgbackrest config for pg2-host
pg2-host=172.18.0.1
etcd addr for host2
ETCD_INITIAL_CLUSTER="db1=https://10.103.0.10:2380,db2=https://172.18.0.1:2380"
patronictl list

2025-10-09 13:19:31,442 - WARNING - Retrying (Retry(total=0, connect=None, read=None, redirect=0, status=None)) after connection broken by 'NewConnectionError('<urllib3.connection.HTTPSConnection object at 0x7ab49ce08b00>: Failed to establish a new connection: [Errno 111] Connection refused')': /version
2025-10-09 13:19:31,442 - ERROR - Failed to get list of machines from https://172.18.0.1:2379/v3beta: MaxRetryError("HTTPSConnectionPool(host='172.18.0.1', port=2379): Max retries exceeded with url: /version (Caused by NewConnectionError('<urllib3.connection.HTTPSConnection object at 0x7ab49cee7290>: Failed to establish a new connection: [Errno 111] Connection refused'))")

Expected behavior

I've set up a cluster and decided to monitor it using Docker containers. But when I run the configuration playbook, I encounter a lot of nuances when reconfiguring. I'd also like to specify a different interface for each keepalived instance, as I might run into a situation like mine, where I have different local interfaces on different nodes on each server.

Steps to reproduce

1. install cluster. (i have two interfaces eth0 eth1)
2. add docker
3. reconfigure cluster via command line

TASK [vitabaks.autobase.bind_address : Get docker0 interface IP] ***************************************************************************************************************
ok: [10.103.0.10]
ok: [10.103.0.20]

Installation method

Command line

System info

ubuntu 24
autobase 2.3.2

Additional info

I can fix a lot of problems use this paratmetrs:

          ansible_host: 10.103.0.10
          postgresql_exists: true
          patroni_bind_address: 10.103.0.10
          etcd_bind_address: 10.103.0.10

[vitabaks]

If your servers have multiple network interfaces, we recommend not relying on automatic private IP detection.
Instead, explicitly define the correct IP address for each host using the bind_address variable in your Ansible inventory.

This ensures that Patroni and other cluster components use the intended network interface.

[vitabaks]

I'd also like to specify a different interface for each keepalived instance

use the vip_interface variable. You can specify an individual value for each host in the inventory.