| Re: VE connect with physical interface [message #23278 is a reply to message #23247] |
Thu, 15 November 2007 08:19   |
khorenko
Messages: 533
Registered: January 2006
Location: Moscow, Russia
|
Senior Member |
|
|
Great, thank you for the info.
The situation is following:
VE1 is pinging VE2 (10.10.10.3).
Routing table on a VE1 says that the packet should go through the venet0 interface (as no any other interfaces exists inside a VE1 :) ). As in any other Linux node the source IP for such a packet is set as a FIRST IP from the interface where this packet is routed.
In your case the first IP is 143.69.77.8 (which can also be seen in tcpdump on a HN2 on bond0 - the packet source IP is 143.69.77.8).
VE2 receives the packet and sends reply to the 143.69.77.8, and HN2 has a route rule that sends such a packet to the eth2. That's it.
So the problem is: VE1 sets "incorrect" (not that as expected) source IP while pinging VE2 (10.10.10.3). This can be handled by adding a route rule in VE1 like:
ip r a 10.10.10.0/24 dev venet0 scope link src 10.10.10.5
You can also take a look at http://kb.swsoft.com/en/3061
Hope this helps.
--
Konstantin.
If your problem is solved - please, report it!
It's even more important than reporting the problem itself...
|
|
|
|
OpenVZ Forum
Members
Search
Help
Register
Login
Home
