Description

Maybe Docker/Moby should start supporting IPv6.

And by 'supporting' I don't mean having to skim through GitHub comments and the Docker forum for hours to enable basic functionality.

I tried to enable IPv6 for user-defined bridge networks.

To enable user-defined bridge networks, default-address-pools must be set. I set default-address-pools, causing Docker to not start anymore.

There's this months old issue that says an IPv4 address address pool should be before placed before an IPv6 address pool.

Aside from the code smell that this bug brings, this requires me to set an IPv4 address pool. Why?

I attempted to remove fixed-cidr-v6, as I will only be using user-created networks, which caused Docker not to start anymore.

I found this forum post, which says:

The subnet specified in fixed-cidr-v6 defines what the docker0 network on the machine should look like. No more, no less. I still think this is poorly documented.

And indeed, the documentation says that IPv6 cannot be disabled for the default bridge. However, nowhere does the documentation say that fixed-cidr-v6 is the subnet used for the docker0 bridge.

To add the IPv4 address pool (which should not be needed), I need to know what the IPv4 address pool is.

This is not documented anywhere. I found this almost 4 year old issue requesting for this to be documented.

Finally having gotten Docker to stop complaining about missing or misordered settings, Docker crashed immediately on startup because it used all available RAM.

I found this almost 4 year old issue, which says that at least an /80 should be used. Makes sense, but why does Docker crash because of this 4! years after the bug report?!