Docker removes host ipv6 default route

Open Source Projects DockerEngine
1

with ipv6 enabled Docker seems to remove the ipv6 default route from thehost on startup.
The default route is created by slaac

OS CentOS Linux release 7.7.1908 (Core) running under VMware

docerk version
docker-ce-19.03.3-3.el7.x86_64
docker-ce-cli-19.03.3-3.el7.x86_64
containerd.io-1.2.6-3.3.el7.x86_64

daemon.json file
{
“bip”: “192.168.0.1/16”,
“fixed-cidr”: “192.168.0.0/16”,
“fixed-cidr-v6”: “fdaa:bbbb:cccc::/48”,
“ipv6”: true,
“debug”: true
}

Recreated by disabling docker service, rebooted host and default route appears in ip -6 route command output,
start docker using service docker start
default route disappears from ip -5 route output

before

unreachable ::/96 dev lo metric 1024 error -113 pref medium
unreachable ::ffff:0.0.0.0/96 dev lo metric 1024 error -113 pref medium
2001:xxx:xxxx:40::/64 dev ensxxx proto kernel metric 256 expires 2591980sec pref medium
unreachable xxxx:a00::/24 dev lo metric 1024 error -113 pref medium
unreachable xxxx:7f00::/24 dev lo metric 1024 error -113 pref medium
unreachable xxxx:a9fe::/32 dev lo metric 1024 error -113 pref medium
unreachable xxxx:ac10::/28 dev lo metric 1024 error -113 pref medium
unreachable xxxx:c0a8::/32 dev lo metric 1024 error -113 pref medium
unreachable xxxx:e000::/19 dev lo metric 1024 error -113 pref medium
unreachable xxxx:ffff::/32 dev lo metric 1024 error -113 pref medium
fe80::/64 dev ensxxx proto kernel metric 256 pref medium
default via fe80::xxx:xxxx:xxxx:a8 dev ensxx proto ra metric 1024 expires 1780sec hoplimit 64 pref medium

after
unreachable ::/96 dev lo metric 1024 error -113 pref medium
unreachable ::ffff:0.0.0.0/96 dev lo metric 1024 error -113 pref medium
2001:xxx:xxxx:40::/64 dev ensxxx proto kernel metric 256 expires 2591980sec pref medium
unreachable xxxx:a00::/24 dev lo metric 1024 error -113 pref medium
unreachable xxxx:7f00::/24 dev lo metric 1024 error -113 pref medium
unreachable xxxx:a9fe::/32 dev lo metric 1024 error -113 pref medium
unreachable xxxx:ac10::/28 dev lo metric 1024 error -113 pref medium
unreachable xxxx:c0a8::/32 dev lo metric 1024 error -113 pref medium
unreachable xxxx:e000::/19 dev lo metric 1024 error -113 pref medium
unreachable xxxx:ffff::/32 dev lo metric 1024 error -113 pref medium
fe80::/64 dev ensxxx proto kernel metric 256 pref medium
fdaa:bbbb:cccc::/48 dev docker0 proto kernel metric 256 pref medium
fdaa:bbbb:cccc::/48 dev docker0 metric 1024 pref medium
fe80::/64 dev ensxxx proto kernel metric 256 pref medium
fe80::/64 dev docker0 proto kernel metric 256 pref medium
fe80::/64 dev vethae461bc proto kernel metric 256 pref medium
fe80::/64 dev veth235296c proto kernel metric 256 pref medium
fe80::/64 dev veth8d8c6dc proto kernel metric 256 pref medium
fe80::/64 dev veth532e9cc proto kernel metric 256 pref medium

2

There are some known issues with libnetwork in engine 19.03.3. See https://github.com/docker/for-linux/issues/810 and https://github.com/docker/docker-ce/releases. While not directly related to your problem symptoms, it’s possible there may be some unexpected interactions there.

It may be worth a moment to see how the latest revision, 19.03.4, does in your use case.

3

I downloaded
docker-ce-19.03.4-3.el7.x86_64.rpm
and
docker-ce-cli-19.03.4-3.el7.x86_64.rpm
from the stable repo with the same result where the ipv6 defualt route is deleted.