Hello everyone,

 

I'm on unraid 6.11.5. I have ipv6 disabled in my network configuration, but when I do ip addr show, I see multiple ipv6 in local-link. Problem is when I want to access samba share \\servername, it return a fe80 addr that I can't access (even when on same switch).

 

ip addr show (part of it)

network configuration

 

is there a way to not have ipv6? I tried enabling ipv6, the server received and address but I still receive the fe80 addr when I ping (and I can't find where this address is because I don't have any ipv6 dns).

 

Is there a way to completely disable ipv6?

 

Thank you

When IPv4 only is used all the "regular" interfaces have no IPv6 assignments at all

Example

217: br0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
    link/ether a0:36:9f:79:84:ac brd ff:ff:ff:ff:ff:ff
    inet 10.0.101.13/24 scope global br0
       valid_lft forever preferred_lft forever
218: br0.2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
    link/ether a0:36:9f:79:84:ac brd ff:ff:ff:ff:ff:ff
219: br0.3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
    link/ether a0:36:9f:79:84:ac brd ff:ff:ff:ff:ff:ff
220: br0.4: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
    link/ether a0:36:9f:79:84:ac brd ff:ff:ff:ff:ff:ff
221: br0.5: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
    link/ether a0:36:9f:79:84:ac brd ff:ff:ff:ff:ff:ff
222: br0.6: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
224: br1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
    link/ether a0:36:9f:79:84:ad brd ff:ff:ff:ff:ff:ff
    inet 10.0.101.84/24 brd 10.0.101.255 scope global dynamic noprefixroute br1
       valid_lft 85402sec preferred_lft 74602sec

 

In your picture the IPv6 assignments are to virtual interfaces (veth).

Make sure your docker configuration and docker containers are not configured for ipv6.

 

As a test disable the docker and vm services and check again.

 

I don't see anything related to ipv6 in this.

 

 

Could it be because I've put ipv6 once, it's now stuck with it?

 

I'm also seeing ipv6 on other interface after docker is down

Edited May 2, 20233 yr by Nodiaque

Your "regular" interface is br0, which is ipv4 only

 

Try the following:

- Re-enable IPv6 for br0 (docker and vm services must be stopped to do this)

- Go to docker configuration and if any IPv6 assignment is checked, uncheck it

- Reconfigure br0 back to ipv4 only again

- Start docker and vm services

 

No ipv6 in docker:

 

and network setting has ipv4 + 6 enabled

if I try to ping my server, something I get ipv4 other time ipv6 fe80

 

I seems to get ip from br-22a....

 

If I turn back ipv4 only (I also disabled bonding cause I don't use it anyway)

 

 

still have some int6 ip, docker isn't running nor vm

 

edit: but when I ping, I always get ipv4 now. So, could it be a specific docker guest that's doing that?

Edited May 2, 20233 yr by Nodiaque

br-22a634b77f67 is created by a docker container, you need to find out which container

 

but docker is disabled

 

edit: Oh! could it be my custom ducker network? But docker is stopped so it shouldn't be there like we see in the network tab?

Edited May 2, 20233 yr by Nodiaque

A container created this interface, it won't go away when you stop docker.

 

If you can't find out, reboot your system but don't autostart docker containers, that should clear it.

 

 

ok, if I restart unraid and prevent all container from starting up?

1 minute ago, Nodiaque said:

ok, if I restart unraid and prevent all container from starting up?

 

We were typing at the same time....

Yes you should do that.

 

ok, so right after restart, I'm like this with docker daemon down

 

 

I start the docker daemon, no docker is running

Here the docker network ls

 

now ping randomly reply either ipv4 or ipv6, but the ipv6 work

 

the only one that have ipv6 right now is the shim-br0

 

Edit: I've ran docker network inspect on all docker network, they all have enableipv6: false

 

I didn't start any docker yet

 

Edited May 2, 20233 yr by Nodiaque

8 minutes ago, Nodiaque said:

the only one that have ipv6 right now is the shim-br0

 

The shim-br0 interface is created because host access to custom network is enabled.

If you don't need host access, you should disable it under Docker settings

 

I put on my todo list to include shim interfaces as well when not using ipv6. Merci  

 

You can manually disable ipv6 for this interface by doing

echo 1 > /proc/sys/net/ipv6/conf/shim-br0/disable_ipv6

 

Ah, no I need it, specially since my DNS run on docker (and many db and stuff). When I disable that, it crash many stuff.

 

I tried starting 1 by 1 docker, 2 started and already found a problem.

 

1rst docker that started was on br0, which changed nothing in the output of the command.

The 2nd docker is on my-bridge, which is the br-22a.... It added a new entry (#12) with ipv6 and it forced br-22... to get ipv6.

 

I'll try with the shim ipv6 off see if it fix

2 minutes ago, Nodiaque said:

The 2nd docker is on my-bridge

 

You need to delete and recreate this network "'my-bridge" without ipv6 settings.

 

but it already have enableipv6: false

 

  "Name": "my-bridge",
        "Id": "22a634b77f67e5e8415aecbd20ff05b0c784f4ff29d740bc2b365068745c3eda",
        "Created": "2023-02-10T18:42:56.546787388-05:00",
        "Scope": "local",
        "Driver": "bridge",
        "EnableIPv6": false,
        "IPAM": {
            "Driver": "default",
            "Options": {},
            "Config": [
                {
                    "Subnet": "192.168.255.0/24",
                    "Gateway": "192.168.255.1"
                }
            ]
        },
        "Internal": false,
        "Attachable": false,
        "Ingress": false,
        "ConfigFrom": {
            "Network": ""
        },
        "ConfigOnly": false,
        "Containers": {},
        "Options": {},
        "Labels": {}

 

3 minutes ago, Nodiaque said:

but it already have enableipv6: false

 

Yes config looks alright, docker apparently doesn't fully disable ipv6 and always create the link-local address

 

I did a quick test,

docker network create my-bridge --subnet=192.168.255.0/24 --gateway=192.168.255.1

 

# docker network inspect my-bridge
[
    {
        "Name": "my-bridge",
        "Id": "a92c67493cc6ae2dc42c23b538c6d2eae3faf0498b456a017edddad091a3d8ed",
        "Created": "2023-05-02T19:30:31.173992897+02:00",
        "Scope": "local",
        "Driver": "bridge",
        "EnableIPv6": false,
        "IPAM": {
            "Driver": "default",
            "Options": {},
            "Config": [
                {
                    "Subnet": "192.168.255.0/24",
                    "Gateway": "192.168.255.1"
                }
            ]
        },
        "Internal": false,
        "Attachable": false,
        "Ingress": false,
        "ConfigFrom": {
            "Network": ""
        },
        "ConfigOnly": false,
        "Containers": {},
        "Options": {},
        "Labels": {}
    }
]

 

 

268: br-a92c67493cc6: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN group default
    link/ether 02:42:9a:6e:fa:95 brd ff:ff:ff:ff:ff:ff
    inet 192.168.255.1/24 brd 192.168.255.255 scope global br-a92c67493cc6
       valid_lft forever preferred_lft forever

 

No IPv6 address.

 

I am using Unraid 6.12.0-rc5 with docker version 20.10.23

 

I don't have any ipv6 address until a docker start on it, that's the weird thing. I'll see with shim disabled, I'm currently rebooting to erase all trace of it

Hmmm... do I need to put the command you gave me un a script for each reboot? Because I still have ipv6 on shim. I also see ipv6 on loopback

 

 

This is after restart and starting docker daemon, but no docker. We can see the custom bridge doesn't have IPv6 yet

hmmm... I turned off docker and decided to go look in the shim place. Shim isn't there but found br-2.... and:

There's also a all folder, can I simply change this one to 1 (because it's also at 0)?

Edited May 2, 20233 yr by Nodiaque

1 minute ago, Nodiaque said:

Hmmm... do I need to put the command you gave me un a script for each reboot?

 

The command I gave is a temporary fixe and does not survive a reboot (you need to run it again)

 

1 minute ago, Nodiaque said:

I also see ipv6 on loopback

 

That is correct, loopback interface is configured for both ipv4 and ipv6. It won't interfere with docker

 

1 minute ago, bonienl said:

 

The command I gave is a temporary fixe and does not survive a reboot (you need to run it again)

 

Ok, problem is when I run it before docker, it says path not found cause shim-br0 doesn,t exist. If I create it after docker is ran, it's too late. Then when I turn off docker, it's gone.

 

Also, all other interface (including default and all) have 0 in disable_ipv6. Shouldn't they all have 1 since I disabled ipv6?

2 minutes ago, Nodiaque said:

Shim isn't there

When docker is stopped all the shim interfaces are removed, they only exist when docker is running

 

3 minutes ago, Nodiaque said:

can I simply change this one to 1

You can, but look at my test with Unraid 6.12.0-rc5

 

yeah I understand with 6.12, I'm on 6.11. Might be something different in 6.12? Because as you can see, mine doesn't work. When I start docker and put shim 1, it still has a ipv6

 

edit: I put it in all and it worked, shim doesn't have ipv6. Now the problem is how I make this permanent accross reboot because even user script won't be fast enough

 

also, I must test booting these vm and see if it get one

Edited May 2, 20233 yr by Nodiaque

It worked when I put all disable_ipv6 = 1. Even loopback doesn't have inet6 anymore and I started the vm that "spawned" the inet6 on the br-2 earlier and nothing

 

 

Might be a 6.11.5 bug? Would be good when someone select ipv4 only to really disable it everywhere like in all.