Network Problems After Installation from Docker

[Doenerknall]

I am having strange behaviour after installing 7.3-3 from scratch on Lenovo Thinkcentre M900.
After installing docker in a LXC or VM i get kicked out of the LXC.
I can reproduce that with different distros. First i tried Ubuntu and with Rockylinux the same.
I am logged in via ssh and get kicked out with "client_loop: send disconnect: Broken pipe". When retrying to relogin , "ssh: connect to host 10.0.10.120 port 22: Connection refused". And now comes the fun, after nmap -sS 10.0.10.120 i am able to relogin.
Nested and keyctl is hocked.
See attached screenshot. Can someone explain that to me?
Ok, same problem with 7.2-3. What am i doing wrong?

 

[Hannes Laimer]

Hey,

I assume you mean after starting docker containers or is this happening right after the installation of docker? Could you post the output of journalctl -u sshd.service -b and lsof -i -P -n | grep LISTEN after it kicked you out(and you're connected again)?

 

[Doenerknall]

Hey,

I assume you mean after starting docker containers or is this happening right after the installation of docker? Could you post the output of journalctl -u sshd.service -b and lsof -i -P -n | grep LISTEN after it kicked you out(and you're connected again)?

This happens after installing docker and setting up container. At the moment i am on vacation. On friday i am back and can post the output. Without docker installed the LXC container run without this problems. The other Problem is, before this happens DNS lookups ok, after reconnecting to the LXC container i only get nslookup after the 3 try. Udo

 

[Doenerknall]

I apologize for the confusion. I started with Proxmox from scratch.
Only installed Ubuntu 22.10 LXC container, no docker. Sometimes i can't login to the LXC Container.

snipped of my /etc/sysctl.conf
#disable ipv6

net.ipv6.conf.all.disable_ipv6 = 1
net.ipv6.conf.default.disable_ipv6 = 1
net.ipv6.conf.lo.disable_ipv6 = 1

why is sshd listen on ipv6 address?

On the attached screenshot you can see the wanted output of journactl and lsof.

 

[Doenerknall]

I apologize for the confusion. I started with Proxmox from scratch.
Only installed Ubuntu 22.10 LXC container, no docker. Sometimes i can't login to the LXC Container.

snipped of my /etc/sysctl.conf
#disable ipv6

net.ipv6.conf.all.disable_ipv6 = 1
net.ipv6.conf.default.disable_ipv6 = 1
net.ipv6.conf.lo.disable_ipv6 = 1

why is sshd listen on ipv6 address?

On the attached screenshot you can see the wanted output of journactl and lsof.

Now it happened again:

udo@ubuntu-test:~$ client_loop: send disconnect: Broken pipe

Connected via console from Proxmox:
udo@ubuntu-test:~$ sudo lsof -i -P -n | grep LISTEN
[sudo] password for udo:
systemd 1 root 39u IPv6 51195 0t0 TCP *:22 (LISTEN)
systemd-r 96 systemd-resolve 14u IPv4 51188 0t0 TCP 127.0.0.53:53 (LISTEN)
systemd-r 96 systemd-resolve 16u IPv4 51190 0t0 TCP 127.0.0.54:53 (LISTEN)
sshd 121 root 3u IPv6 51195 0t0 TCP *:22 (LISTEN)
master 273 root 13u IPv4 51764 0t0 TCP 127.0.0.1:25 (LISTEN)
udo@ubuntu-test:~$ sudo journalctl -u sshd.service -b
-- No entries --
udo@ubuntu-test:~$ date
Fri Dec 9 11:56:29 UTC 2022
udo@ubuntu-test:~$ sudo ping google.de
ping: google.de: Temporary failure in name resolution
udo@ubuntu-test:~$ sudo ping google.de
ping: google.de: Temporary failure in name resolution
udo@ubuntu-test:~$ sudo ping google.de
PING google.de (142.251.36.195) 56(84) bytes of data.
64 bytes from muc12s12-in-f3.1e100.net (142.251.36.195): icmp_seq=1 ttl=117 time=31.0 ms

 

[Doenerknall]

Beginner's mistakes, duplicate IPs and old ARP cache entries.
We can close this thread or mark it as solved.