Cold-standby backup

DaLLLi

I’m trying to do a cold-standby backup. An error is generated in the process. Has anyone had such a problem?

DaLLLi

The problem may exist in older installations. There was no network setting in docker-compose.override.yml for the traefik_certdumper container. By default, docker creates a new network named mailcowdockerized_default. The script enumerates networks named mailcowdockerized and looks for containers that are not there.
Solution:
A line should be added for traefik_certdumper.

network_mode: none

ecdlguy

Hi! My installation is not old but I also have the “mailcowdockerized_default”-network:
root@mail:/opt/mailcow-dockerized# docker network ls

NETWORK ID     NAME                                DRIVER    SCOPE
3eabc1ed38dd   bridge                              bridge    local
18399acd6115   docker-mailman_mailman              bridge    local
9ab2dfecd136   host                                host      local
c885340e187c   mailcowdockerized_default           bridge    local
8b0dccf43aaf   mailcowdockerized_mailcow-network   bridge    local
28150a8871b7   none                                null      local

However, no containers are attached to the network:

root@mail:/opt/mailcow-dockerized# docker network inspect mailcowdockerized_default | grep Containers
        "Containers": {},

The default network seems to appear while or after updating to 2025-09b:

root@mail:/opt/mailcow-dockerized# docker network inspect mailcowdockerized_default | grep Created
        "Created": "2025-09-21T21:30:47.068403635+02:00",

This also was the first time create cold standby didn’t work. I don’t understand why this network is created at all. Is it safe to just remove the network?

cheers, Thorsten

DaLLLi

I recently migrated to a new server with new settings. Currently, the only valid network is
mailcowdockerized_mailcow-network
I think you can safely remove the unnecessary mailcow network and check if the backup is running.

ecdlguy

Ok, any idea from where the default network is coming from?

stefan21

No idea.

But a simple “docker system prune” will solve this.

ecdlguy

I’ve just updated to version 2025-09c, and the default network has reappeared. Since no containers are using this network, I was wondering why it exists at all — and I think I’ve found the reason.

Because I don’t use IPv6, I previously added the following to docker-compose.override.yml (based on the documentation):

services:
  ipv6nat-mailcow:
    image: bash:latest
    restart: "no"
    entrypoint: ["echo", "ipv6nat disabled in compose.override.yml"]

Since no network is defined for this service, Docker automatically creates a default network which gets attached to it. However, the container exits right after running echo, so the network remains, even though no container actually uses it after startup.

Given that I’m now running a newer version of mailcow, can I safely remove this service from docker-compose.override.yml?
Should I also revert the other changes recommended in the old IPv6 disable instructions?