| Accessing eth devices from HN after being dedicated by netdev_add [message #39807] |
Fri, 04 June 2010 13:57 |
jeronimo
Messages: 4
Registered: August 2009
Location: Czech Republic
|
Junior Member |
|
|
Dear experts, friends,
I'd like to ask you for an advice about a bit unusual setup problem:
I have a HN (host node) having eth0 device and several VLAN devices (e.g., eth0.30 , eth0.40, etc.). I need to dedicate these VLAN devices to VPSs via netdev_add like this:
# vconfig add eth0 30
# vzctl set XXX --netdev_add eth0.30
Up to here everything works pretty fine -- the VPS is able to see the device, setup it and communicate via it. And, if necessary, create further VLANs inside it...
However, I need to solve one another issue: traffic control via tc on the HN. Once I dedicate the device, I'd like to control the amount of traffic (for each device) which it can consume. But, once the device is dedicated via netdev_add, it disappears from the HN and the tc command says:
# tc qdisc add dev eth0.30 ...
Cannot find device "eth0.30"
Thus, I'd like to ask you: is there a way I can do a traffic shaping from the HN, once the devices I want to control had been dedicated to VPSs? Or, is there another solution which can let the VPSs to have "absolute" power on the interfaces' setup (like creation of further VLANs, MAC address changes, IP address changes, etc.)? Or, is there a setup which is feasible and is close to my ideal setup described?
Thank you all for your ideas and remarks.
--best
Tom.
PS: My setup is: Debian 5.0 with 2.6.26-2-openvz-amd64 kernel, vzctl version 3.0.23. The VPSs do run Debian 5.0 as well...
PS2: Maybe, it might be possible to do the shaping on the eth0 device, but I cannot imagine how the dedicated VLAN devices can be recognized/identified there, since they can have arbitrary MAC or IP -- just the master VLAN ID (e.g., 30) cannot be changed by VPS at all...
|
|
|
|
OpenVZ Forum
Members
Search
Help
Register
Login
Home
