OpenVZ Forum


Home » General » Support » Multiple veth in VE assigned to proper vmbr on HN? (I need two bridged nics inside a container each on a separate physical host network)
Multiple veth in VE assigned to proper vmbr on HN? [message #39940] Mon, 28 June 2010 20:13 Go to next message
tnelson is currently offline  tnelson
Messages: 3
Registered: June 2010
Junior Member
Greetings all-

I posted this to the list last week but there is very little (almost none!) traffic. Sorry for the crosspost...

I'm having issues with what I would think is a common scenario. I have a host node running CentOS 5.0 with the OpenVZ kernel (2.6.18-164.11.1.el5.028stab068.3) and appropriate tools. OpenVZ works just fine. The host node has two physical network interfaces, one which resides on a public subnet routed to me by the ISP, and another which is on an internal private network. The HN needs to be accessible via the public subnet (default gateway is ISP router), and still have access to internal network resources, namely NFS. This works just fine on the HN.

I've been attempting to get this same functionality inside of the VE's but using venet has been an absolute nightmare. Traffic seems to be sourced from the public interface regardless of destination and there are other 'oddities' which I don't particularly care for.

I would like to have the same public/private functionality available to the VE's but using veth. The VE's should have two veth interfaces (eth0 and eth1 I would assume) with each being 'mapped' to the appropriate interface on the HN for public/private network connectivity. On the HN I've created two bridges, vmbr0 and vmbr1. vmbr0 includes eth0 which is for public connectivity. vmbr1 includes eth1 which is for private connectivity.

The big question is how do I configure the above scenario? How do I assign multiple veths to each VE and ensure that veth is on the proper vmbr on the HN?

All suggestions welcome. Thanks!

--Tim
icon3.gif  Re: Multiple veth in VE assigned to proper vmbr on HN? [message #39941 is a reply to message #39940] Mon, 28 June 2010 21:09 Go to previous message
tnelson is currently offline  tnelson
Messages: 3
Registered: June 2010
Junior Member
For the archives:

After some additional testing, I found what works for me. As noted in my OP, on the host node eth0 in vmbr0 and eth1 in vmbr1:

vzctl set 102 --netif_add eth0,,,,vmbr0 --save
vzctl set 102 --netif_add eth1,,,,vmbr1 --save
vzctl restart 102

As per the wiki article, the above syntax works on vzctl > 3.0.22 but ymmv on lesser versions.

For posterity, I'll cross post this solution the mailing list.
Previous Topic: Hardware Failure with new kernel
Next Topic: Kmemsize - Always failing and super huge
Goto Forum:
  


Current Time: Fri Sep 27 14:23:32 GMT 2024

Total time taken to generate the page: 0.04245 seconds