Configure eth0 on a new container

[bdunbar]

What's the best method to configure eth0 for a new container? It's easy enough to do this via GUI but I really want to do this by script.

So far I can create nodes using pvesh. Assuming I want a container ID of '101'

pvesh create /nodes/prox_host_1/openvz -vmid 101 -hostname bob (redacted for clarity)

I then add eth0 using vzctl

vzctl set 101 --netif_add eth0,,,,vmbr0 --save

Next IP address ...

vzctl set 101--ipadd 10.10.3.101 --nameserver 8.8.8.8


Issues

This doesn't seem to populate the container's /etc/network/interfaces. Am I supposed to write directly to /var/lib/vz/root/101/etc/network/interfaces?

When I tell it to use parameter --gateway I'm told 'unrecognized option'. So how DO I add a gateway?

I feel like I'm so _close_ to getting this thing understood but I'm missing something obvious.

 

[screenie]

seems you use venet interfaces - in this case you only can assign an ip address from the host interfaces subnet;
with the use of veth (bridged mode) you can use networking inside the container as on a regular pc;

i use only veth interfaces with my containers and thats how it looks like in the config:
NETIF="ifname=eth0,bridge=vmbr0,mac=BED6:23:19:42,host_ifname=veth121.0,host_mac=00:18:513:67:E8"

in your script you need then generate for each container the /etc/network/interfaces file and copy it into the container

 

[bdunbar]

Excellent! Thank you.

FWIW my process is
script as outlined in post
script uses predefined IP/gateway/netmask info to build a temp file - /tmp/scriptname.txt
script executes scp to copy /tmp/scriptname.txt > hypervisor:/var/lib/vz/root/${VMID}/etc/network/interfaces
script reboots container using vzctl restart

I've tested this on a node, seems to work clean.

Works, I'm happy, now on to building stuff.

If there is a better / cleaner / elegant way to get this up and running, I'd be interested!