openstack ansible deployment fail after creating lxc containers while running apt-get update command inside container

Question

The openstack ansible created containers where running but they don't have access to internet. When I manually add routing then I get following error:

Err:1 http://archive.ubuntu.com/ubuntu bionic InRelease
  Clearsigned file isn't valid, got 'NOSPLIT' (does the network require authentication?)

# lxc-ls -f
NAME                                STATE   AUTOSTART GROUPS            IPV4                         IPV6 UNPRIVILEGED
infra1_galera_container-40eb8c2c    RUNNING 1         onboot, openstack 192.168.56.232               -    false
infra1_glance_container-fe32f950    RUNNING 1         onboot, openstack 192.168.56.194, 10.200.75.51 -    false
infra1_heat_api_container-fcdda89c  RUNNING 1         onboot, openstack 192.168.56.207               -    false
infra1_horizon_container-29382c02   RUNNING 1         onboot, openstack 192.168.56.140               -    false
infra1_keystone_container-c4b4bca5  RUNNING 1         onboot, openstack 192.168.56.61                -    false
infra1_memcached_container-667ec859 RUNNING 1         onboot, openstack 192.168.56.183               -    false
infra1_nova_api_container-6dcd2a9b  RUNNING 1         onboot, openstack 192.168.56.114               -    false
infra1_placement_container-d9a47077 RUNNING 1         onboot, openstack 192.168.56.218               -    false
infra1_rabbit_mq_container-34797cf1 RUNNING 1         onboot, openstack 192.168.56.210               -    false
infra1_repo_container-02a62c03      RUNNING 1         onboot, openstack 192.168.56.94                -    false
infra1_utility_container-a23621d0   RUNNING 1         onboot, openstack 192.168.56.90                -    false

also thing is my lxc bridge has some different network & that is not getting assigned

lxcbr0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 10.0.3.1  netmask 255.255.255.0  broadcast 10.0.3.255

my openstack_user_config.yml defines all infra components on 192.168.56.0/24 network.

Why by default when I run the playbook I don't get internet access on containers? did I missed anything?

Answer 1

Internet access to lxc containers is provided through lxcbr0. Each container should get an IP 10.0.3.0/24 subnet assigned to eth0 (default interface). I have the same issue, which is caused by dnsmasq not handling DHCP request sent to lxcbr0. Restarting dnsmasq works. Following are the steps I took:

Check if DHCP requests are handled with tcpdump on controller host:

tcpdump -ntv -i lxcbr0

Restart dnsmasq:

ps aux | grep dnsmasq
# note pid and copy dnsmasq command
kill dnsmasq-pid
rerun dnsmasq command

or with systemctl start lxc-dnsmasq.service

Answer 2

There should be a need to check the /etc/resolv.conf

The correct nameserver is not added there because this under container apt update is failed.