Docker | docker0 has IP (is accessable from external) only at first run - docker

Question

When I start a simple docker container (e.g. Portainer) with
docker run -d --name portainer -p 9000:9000 -v /var/run/docker.sock:/var/run/docker.sock portainer/portainer
the container is accessable from the internet as expected.
When I stop (docker stop portainer) and start (docker start portainer) the container, the port 9000 is open again (verified with nmap), but the web interface of portainer loads forever.
# first run
networkctl
IDX LINK TYPE OPERATIONAL SETUP
1 lo loopback carrier configured
2 enp35s0 ether routable configured
3 enp36s0 ether no-carrier configuring
5 br-1815f2210327 bridge no-carrier configuring
6 br-7f9b2f2637a1 bridge no-carrier configuring
7 br-a9ae27884558 bridge no-carrier configuring
6552 br-39aac8ad8ef3 bridge routable configuring
6559 docker0 bridge no-carrier configuring
# next run
networkctl
IDX LINK TYPE OPERATIONAL SETUP
1 lo loopback carrier configured
2 enp35s0 ether routable configured
3 enp36s0 ether no-carrier configuring
5 br-1815f2210327 bridge no-carrier configuring
6 br-7f9b2f2637a1 bridge no-carrier configuring
7 br-a9ae27884558 bridge no-carrier configuring
6552 br-39aac8ad8ef3 bridge no-carrier configuring
6559 docker0 bridge no-carrier configuring
I already tried different workarounds that I found on the internet, like
nano /etc/docker/daemon.json
{ "debug": true, "bip": "172.20.0.1/16" }
and this config file in various configurations
nano /etc/systemd/network/docker0.network
#[Match]
#Name=docker0
#[Network]
#IPForward=yes
#[Network]
#Address=172.17.0.1/16
#[Link]
#Unmanaged=yes
(Currently everythings is commented out.)
When I restart the docker daemon with
systemctl restart docker
and then start the docker container
docker start portainer
it's working fine again.
My system is a linux root server hosted by strato.de:
docker -v
Docker version 20.10.6, build 370c289
cat /etc/issue
Ubuntu 20.04.2 LTS
uname -r
5.4.0-73-generic
The problem occurs with all of my docker containers on that server.
I would be very grateful for any further tips.
UPDATE
Docker on Ubuntu doesn't connect to localhost
The mentioned solution seems not work on my server with Ubuntu 20.04.
Yesterday I installed the same OS and docker version in a VM. Everything is working fine there.
Kind regards,
K1LLUM1N471

Answer 1

My problem was that i could ping google only once in a container (docker run --rm alpine ping google.com), after exit it would not ping the next time i ran the same command. In ifconfig docker0 the inet address was gone after exiting the container the inet6 was still there after running the command once.
when running networkctl status the docker0 link is at configuring.
This might do the trick:
the default Netplan config (/etc/netplan/01-netcfg.yaml)in my Ubuntu 22.04 server from Strato (dedicated server) is:
network:
version: 2
ethernets:
mainif:
match:
name: '*'
dhcp4: yes
replace it with something like this:
network:
version: 2
ethernets:
enp3s0:
dhcp4: yes
dhcp6: no
enp2s0:
dhcp4: yes
dhcp6: no
apply netplan config
sudo netplan try or sudo netplan apply
restart Docker
sudo systemctl restart docker
when now running networkctl the docker0 link should be unmanaged

Answer 2

For your interrest:
I know that is not the best answer, but in my case I solved the problem by downgrading the OS on the root server :(
cat /etc/issue
Ubuntu 18.04.4 LTS
docker -v
Docker version 20.10.7, build f0df350