1

Quick summary:

We have a modeling/simulation environment (using EXata) with external system in the loop. So, external systems <--simulation environment --> external systems. Communication from the external systems to the the simulation environment is via Ethernet. So, physical interface to physical interface. The workstation hosting the simulation software binds one of its physical interfaces (sub-interfaces).

On to the problem (might be impossible to solve).

I have one Windows 7 workstation. I need to create multiple sub-interfaces on this workstation. I do this by adding additional IPs and gateways on the physical interface. Each additional IP (or sub-interface) must be on a different subnet and therefore have its own gateway. Windows lets me set this up. On the Linux box hosting the simulation software I create the respective sub-interfaces on its physical interface. So basically I want one Ethernet cable to communicate between the Linux and Windows sub-interfaces.

Now, I can get this to work in a Linux to Linux setup using ip routes and ip rules, but I am having problems with Windows. Unfortunately we must use Windows for this....ugggg

So, in this example:

Windows 7 box:

Main IP: 190.0.1.101 Main GW: 190.0.1.1

Secondary IP: 190.0.2.101 Secondary GW: 190.0.2.1

Linux (Ubuntu 16.04) box:

eno1: 190.0.1.100 eno1:1 190.0.2.100

Simulation binds:

Virtual node 1 to eno1 Virtual node 2 to eno1:1

All gateways exist in the simulation as virtual routers.

First problem: I cannot send any traffic through the secondary IP on the Windows box. If I force a ping through that interface using the -S flag I get a general failure error. Now this works from the Main IP. I am able to ping my virtual router and I do not receive any errors. It just fails on any addition IPs I add to this Windows interface.

Second problem. I want to be able to ping from one Windows sub-interface out to the simulation and return on the other Windows sub-interface. So, I want my traffic to leave the Window box go through the simulation and return to the Windows box even though those IPs are on the same Windows sub-interface. This does not seem to work though, no matter how I set the routes and metrics the traffic will never leave that Windows NIC.

Is this possible???????

  • A diagram would help – Ross Jan 10 '18 at 22:46
0

Each additional IP (or sub-interface) must be on a different subnet and therefore have its own gateway. Windows lets me set this up.

So you were able to find out that with static IP settings, windows allows multiple static IPs with one adapter -- which does not create sub-interfaces.

So basically I want one Ethernet cable to communicate between the Linux and Windows sub-interfaces.

The best suggestion I have is to install a broadcom or intel NIC card that supports 802.1q trunk with VLAN interfaces. I know this already works in Linux so you just need to make it work for windows. Trunk VLANs with VLAN interfaces for both machines.

I only have experience with the INTEL NIC card with mutli-VLANs and you will need to find a NIC card compatible with the INTEL ANS driver.

Linux how-to reminder.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.