[lxc-users] Networking in LXC
Fajar A. Nugraha
list at fajar.net
Tue Jun 10 09:41:16 UTC 2014
On Tue, Jun 10, 2014 at 4:18 PM, Fajar A. Nugraha <list at fajar.net> wrote:
> On Tue, Jun 10, 2014 at 3:12 PM, Ajith Adapa <ajith.adapa at gmail.com> wrote:
>> Hi,
>>
>> First I need to really thank the community for helping me out in
>> starting LXC container on fedora 20.
>>
>> I have some basic questions regarding networking in LXC.
>>
>> 1. Is there any tutorial or doc regarding support for various network
>> options in lxc container ?
>
> Probably http://manpages.ubuntu.com/manpages/trusty/man5/lxc.container.conf.5.html
> ?
>
> IIRC Fedora rawhide has lxc 1.0.3, so if you update to that you should
> have the same manpage. Otherwise you'd still be using lxc-0.9.0 which
> might be missing some features.
>
>>
>> 2. When I login into container and try ifconfig command I am getting
>> error saying "command not found" but I am able to run the same command
>> using lxc-attach. Any reason why ?
>>
>
> incorrect PATH? Try
>
> (from the host) lxc-attach -n CONTAINER_NAME -- echo $PATH
> (inside the container) echo $PATH
>
> in your case those two should display different output
>
>> 3. Is it possible to attach a physical interface to lxc container
>> which is in running state ? Currently we need to set the configuration
>> in the config file and restart the container.
>
> There's probably an easier way. The long way would be like this:
>
> # lxc-start -d -n template
>
> # lxc-info -n template
> Name: template
> State: RUNNING
> PID: 8320 <= this is what we need, the PID of a process
> inside the container
> CPU use: 0.93 seconds
> BlkIO use: 6.28 MiB
> Memory use: 18.72 MiB
> KMem use: 0 bytes
> Link: vethDUGP01
> TX bytes: 1.24 KiB
> RX bytes: 84 bytes
> Total bytes: 1.32 KiB
>
> # mkdir -p /run/netns
>
> # touch /run/netns/8320 <= this one could be any name you want, which
> would then be used by "ip ... netns"
>
> # mount --bind /proc/8320/ns/net /run/netns/8320
>
> # ip link show dummy1
> 8: dummy1: <BROADCAST,NOARP> mtu 1500 qdisc noqueue state DOWN mode
> DEFAULT group default
> link/ether 76:c6:a2:7f:c6:57 brd ff:ff:ff:ff:ff:ff
>
> # ip link set dummy1 netns 8320
>
> # ip link show dummy1
> Device "dummy1" does not exist.
>
> # lxc-attach -n template -- ip link show dummy1
> 8: dummy1: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN mode
> DEFAULT group default
> link/ether 76:c6:a2:7f:c6:57 brd ff:ff:ff:ff:ff:ff
>
... and if you use real network device (e.g. eth1) instead of dummy1
like in my example, you should probably run this as well
# ip netns del 8320 (or whatever the netns name you choose above)
This is needed if you want the interface to be available again in the
host after the container shuts down. Otherwise, even when the
container is shutdown, the network namespace will still remains,
keeping the moved interface in it.
--
Fajar
More information about the lxc-users
mailing list