Hi,
I have a cluster of two machines (hosts), with two ethernet interfaces each.
On Host one, I have the system interface configuration as:
Interface Active Ports Autostart IP Address Subnet Mask Gateway
vmbr0 Yes eth2 Y 10.1.1.1 255.255.255.0
vmbr1 Yes eth1 Y 192.168.100.130 255.255.255.0 192.168.100.254
On Host 2 it is:
Interface Active Ports Autostart IP Address Subnet Mask Gateway
vmbr0 Yes eth2 Y 10.1.1.2 255.255.255.0
vmbr1 Yes eth1 Y 192.168.100.131 255.255.255.0 192.168.100.254
Eth1 is plugged into a 100mbit firewall, and is the access to the internet (Via 1 to 1 NAT)
Eth2 is a 1000mbit cable, directly attached between machines. It is on vmbr0 so it is used for migrations and backups at the gigabit speed.
On Host 1 I have created an openvz container (container 1), and given it two IP addresses, separated by a space in the venet network address field of 192.168.100.132, 10.1.1.3
On Host 2 I have created an openvz container (container 2), and given it two IP addresses, separated by a space in the venet network address field of 192.168.100.133, 10.1.1.4
From container 1, I can ping container two by using its 192.168.100.133 address, but I cant ping it if I use its 10.1.1.4 address (Which has better bandwidth available, and less latency).
If I do a TCPdump on Host 2 I get:
23:31:50.728459 IP 192.168.100.132 > 10.1.1.4: ICMP echo request, id 2051, seq 22, length 64
I figure that my ping request from my container is going out my hosts vmbr1 interface and therefore not getting a reply, because both container IP addresses are bound to the vmbr1 interface.
How do I make it so my 10.1.1.X traffic goes across my hosts vmbr0 interface?
Thanks in advance.
Daniel
I have a cluster of two machines (hosts), with two ethernet interfaces each.
On Host one, I have the system interface configuration as:
Interface Active Ports Autostart IP Address Subnet Mask Gateway
vmbr0 Yes eth2 Y 10.1.1.1 255.255.255.0
vmbr1 Yes eth1 Y 192.168.100.130 255.255.255.0 192.168.100.254
On Host 2 it is:
Interface Active Ports Autostart IP Address Subnet Mask Gateway
vmbr0 Yes eth2 Y 10.1.1.2 255.255.255.0
vmbr1 Yes eth1 Y 192.168.100.131 255.255.255.0 192.168.100.254
Eth1 is plugged into a 100mbit firewall, and is the access to the internet (Via 1 to 1 NAT)
Eth2 is a 1000mbit cable, directly attached between machines. It is on vmbr0 so it is used for migrations and backups at the gigabit speed.
On Host 1 I have created an openvz container (container 1), and given it two IP addresses, separated by a space in the venet network address field of 192.168.100.132, 10.1.1.3
On Host 2 I have created an openvz container (container 2), and given it two IP addresses, separated by a space in the venet network address field of 192.168.100.133, 10.1.1.4
From container 1, I can ping container two by using its 192.168.100.133 address, but I cant ping it if I use its 10.1.1.4 address (Which has better bandwidth available, and less latency).
If I do a TCPdump on Host 2 I get:
23:31:50.728459 IP 192.168.100.132 > 10.1.1.4: ICMP echo request, id 2051, seq 22, length 64
I figure that my ping request from my container is going out my hosts vmbr1 interface and therefore not getting a reply, because both container IP addresses are bound to the vmbr1 interface.
How do I make it so my 10.1.1.X traffic goes across my hosts vmbr0 interface?
Thanks in advance.
Daniel