I'm having some trouble with doing some network latency simulation using a FreeBSD 10.2 VM on ESXi, dummynet, a bridge and a management interface. Essentially I have the following:
I set up the port groups in VMWare in promiscuous mode, and the bridge works as designed prior to implementing any ipfw pipes. Meaning traffic from network host 1 traverses the bridge and is forwarded to network host 2 without issue, and vice versa.
When I create pipes, it works fine too. The problem occurs when I add the delay parameter onto the pipe. When I add this configuration, traffic that would normally be forwarded to network host 2 is instead forwarded out the management network, which never reaches the intended destination.
ipfw configuration is as follows:
Is there some configuration that I'm missing that can force the traffic that has gone through the pipe out a particular interface?
Thanks,
Matt
[EDIT]
I should probably note that IO believe the reason the packets are being sent out the management interface is because they're following the default route. So basically the presence of the pipe action removes the packet from bridge forwarding and it's next-hop is determined by the layer-3 properties of the packet. Since the next hop of a packet grabbed by the pipe is always known, if there was a way to force anything from a pipe out a certain interface, that would solve the problem.
Code:
[network host 1] --- [Port Group 1] ----
|
|
- [FreeBSD 10.2] -------------- [mgmt network]
|
|
[network host 2] --- [Port Group 2] ----I set up the port groups in VMWare in promiscuous mode, and the bridge works as designed prior to implementing any ipfw pipes. Meaning traffic from network host 1 traverses the bridge and is forwarded to network host 2 without issue, and vice versa.
When I create pipes, it works fine too. The problem occurs when I add the delay parameter onto the pipe. When I add this configuration, traffic that would normally be forwarded to network host 2 is instead forwarded out the management network, which never reaches the intended destination.
ipfw configuration is as follows:
Code:
#ipfw 64000 add pipe 1 ip from any to any recv vmx1
#ipfw pipe 1 config delay 10
#Is there some configuration that I'm missing that can force the traffic that has gone through the pipe out a particular interface?
Thanks,
Matt
[EDIT]
I should probably note that IO believe the reason the packets are being sent out the management interface is because they're following the default route. So basically the presence of the pipe action removes the packet from bridge forwarding and it's next-hop is determined by the layer-3 properties of the packet. Since the next hop of a packet grabbed by the pipe is always known, if there was a way to force anything from a pipe out a certain interface, that would solve the problem.