My searches on the forum suggest inputting something in the docker.cfg file on the flash drive. I have found the file, but don't understand the format I need to use to make these settings persistent across OS reboots:

{

"default-address-pools": [

{

"base": "172.23.0.0/16",

"size": 24

}

]

}

Thanks for any advice you can offer!

In case you never figured this out, and for the benefit of other people searching for this: you have to put this in /boot/config/go as in add the following (or equivalent) to that file:

mkdir /etc/docker
cat << EOF > /etc/docker/daemon.json
{
  "default-address-pools": [ 
    { "base": "172.23.0.0/16", "size": 24 }
  ]
}
EOF

That will cause the daemon.json file to be created with the given content on every boot.

Edited January 12Jan 12 by clnhlzmn

On 1/12/2026 at 6:49 AM, clnhlzmn said:

In case you never figured this out, and for the benefit of other people searching for this: you have to put this in /boot/config/go as in add the following (or equivalent) to that file:

mkdir /etc/docker
cat << EOF > /etc/docker/daemon.json
{
  "default-address-pools": [ 
    { "base": "172.23.0.0/16", "size": 24 }
  ]
}
EOF

That will cause the daemon.json file to be created with the given content on every boot.

This doesn't work for me, because there is already something writing this file, maybe the nvidia plugin?

root@HTPC:~# cat /etc/docker/daemon.json

{

"runtimes": {

"nvidia": {

"path": "/usr/bin/nvidia-container-runtime",

"runtimeArgs": []

}

}

}

I’m curious why this is needed/is an issue?

You should not be using the internal container IPs and there are better ways to address what I suspect you’re trying to do.

15 hours ago, MowMdown said:

I’m curious why this is needed/is an issue?

You should not be using the internal container IPs and there are better ways to address what I suspect you’re trying to do.

I encounter this issue because I end up running out of address pools after creating a certain number of various docker stacks, when in turn make their own network. Seems to be something like 30 networks or something.

19 hours ago, Bob Cadenza said:

I encounter this issue because I end up running out of address pools after creating a certain number of various docker stacks, when in turn make their own network. Seems to be something like 30 networks or something.

I dont see how you're running out of 254 addresses on the default network size. Are you running more than 254 separate containers?

furthermore you can create custom docker networks which would then get their own separate entire pool of IPs

for example, look at bridge and proxy-net. these both have 512 addressable IPs per pool

Edited September 30Sep 30 by MowMdown

I too have over 30 stacks with separate networks, and the /16 networks chew up an insane amount of address space. Limiting the networks to /24 or smaller makes the route table much cleaner. Since the configuration to change from /16 to /24 does not survive a reboot, this script is needed.

We aren't running out of addresses in a subnet. We are running out of SUBNETS. The problem with using a 16 bit mask is that chews up 65,536 addresses per subnet. By using 24 bit masks we get 256 subnets in the space of the original 16 bit mask.

https://en.wikipedia.org/wiki/Subnet

Specifically, the section on Subnet host count.

Attached below are just some of the stack networks on my unraid

4 minutes ago, matthrolf said:

I too have over 30 stacks with separate networks, and the /16 networks chew up an insane amount of address space. Limiting the networks to /24 or smaller makes the route table much cleaner. Since the configuration to change from /16 to /24 does not survive a reboot, this script is needed.

We aren't running out of addresses in a subnet. We are running out of SUBNETS. The problem with using a 16 bit mask is that chews up 65,536 addresses per subnet. By using 24 bit masks we get 256 subnets in the space of the original 16 bit mask.

https://en.wikipedia.org/wiki/Subnet

Specifically, the section on Subnet host count.

Attached below are just some of the stack networks on my unraid

Exactly, what did you end up doing? Adding the /boot/config/go script in the first reply? Do you have nvidia card?

Just now, Bob Cadenza said:

Exactly, what did you end up doing? Adding the /boot/config/go script in the first reply? Do you have nvidia card?

I am not the person who has the nvidia card. The advice posted above solved my problem, and the result is the screenshot above.

Below is an example of using the default settings, and this host has run out of subnets: