Hi,
I'm trying to setup LVS in a transparent proxy mode to route packets within our Edge Cache cluster. I plan to use LVS in one of the destination hash based scheduler modes (preferably, LbLc).
Does anyone have an example for this kind of setup?
I understand LVS master (director) intercepts the packets to any of the dest IP of the real servers in the cluster (via a FW mark on the Router?) and then forwards it to the appropriate host (based on the Dest IP).
I'm trying to understand what happens in the case of the dest host being overloaded. LbLc scheduling says that LVS master can send packets destined to an overloaded host to another less loaded real server, but, can someone explain how the return path works (I'd need to configure LVS in Direct Server Return mode). Which source IP will the return packets from the new real server include? The new real server or the old real server that the client originally sent the packets to?
Would appreciate any guidance or sample config that uses this kind of setup.
Additionally, I'd also like to use LVS + real server on the same node (similar to local node setup, but, using direct IP), Is this possible with LbLc?
Thanks,
Sudheer
_______________________________________________
Please read the documentation before posting - it's available at:
http://www.linuxvirtualserver.org/
LinuxVirtualServer.org mailing list - lvs-users@LinuxVirtualServer.org
Send requests to lvs-users-request@LinuxVirtualServer.org
or go to http://lists.graemef.net/mailman/listinfo/lvs-users
I'm trying to setup LVS in a transparent proxy mode to route packets within our Edge Cache cluster. I plan to use LVS in one of the destination hash based scheduler modes (preferably, LbLc).
Does anyone have an example for this kind of setup?
I understand LVS master (director) intercepts the packets to any of the dest IP of the real servers in the cluster (via a FW mark on the Router?) and then forwards it to the appropriate host (based on the Dest IP).
I'm trying to understand what happens in the case of the dest host being overloaded. LbLc scheduling says that LVS master can send packets destined to an overloaded host to another less loaded real server, but, can someone explain how the return path works (I'd need to configure LVS in Direct Server Return mode). Which source IP will the return packets from the new real server include? The new real server or the old real server that the client originally sent the packets to?
Would appreciate any guidance or sample config that uses this kind of setup.
Additionally, I'd also like to use LVS + real server on the same node (similar to local node setup, but, using direct IP), Is this possible with LbLc?
Thanks,
Sudheer
_______________________________________________
Please read the documentation before posting - it's available at:
http://www.linuxvirtualserver.org/
LinuxVirtualServer.org mailing list - lvs-users@LinuxVirtualServer.org
Send requests to lvs-users-request@LinuxVirtualServer.org
or go to http://lists.graemef.net/mailman/listinfo/lvs-users