We are in an odd situation right now. On a host where LXC containers used to work while running Rocky Linux (RL) 9.0. After upgrading to RL 9.1, all containers still run, but their networking over the bridged interface doesn’t. More details are below. Any help/hints appreciated.
- We have two hosts running LXC containers. Both hosts have a bridged interface, both named br0. The utility brctl shows that br0 on this RL 9.1 host is up and has interfaces assigned to it. The assigned interface is up too.
- This host in trouble is running RL 9.1 (5.14.0-162.23.1.el9_1.x86_64)
- It runs LXC 4.0.12 (via lxc-ls –version)
- The other working host runs CentOS Linux release 7.9.2009 (Core) (3.10.0-1160.66.1.el7.x86_64)
- The working host runs LXC 3.2.1 (via lxc-ls –version)
- The containers on the host running RL 9.1 are not reachable via ping from another host. Despite the fact that all containers and the two hosts use the same gateway, IP 192.168.20.3, on the same broadcast domain 192.168.20.0/24.
- Inside the LXC containers on the RL 9.1 host, it’s possible to ping br0 with its IP address 192.168.20.20
- But if the hostname of the host is used, there is a ‘Temporary failure in name resolution’ error
- In such LXC containers, ping the common gateway’s IP address 192.168.20.3 produces Destination Host Unreachable
- Ditto for ping 188.8.131.52.
- Firewalld is not active
- SELinux is disabled per getenforce
sudo iptables -L,
sudo iptables -L -t nat, and
sudo nft list rulesetshow empty output