Configuration: 20 UML instances on a P4 2.8 Ghz w/ .75 GB memory a bridge (uml-bridge) set up with 20 interfaces, one for each of the UML instances uml-bridge has ip address of 10.3.82.4 UML instances have addresses 10.3.82.5-24 UML instances have a default route of 10.3.82.4 eth0 has address of 10.3.0.82 zebra+rip advertise the 10.3.82.x route as 10.3.0.82 ip_forwarding is enabled
Problem: When a burst of UDP packets is sent to a number of UML hosts (i.e. several packets to each host), packets are lost between eth0 and uml-bridge. tcpdump on eth0 shows all incoming packets. tcpdump on uml-bridge shows missing packets. Running ifconfig for uml-bridge shows lost TX packets at about the same rate as is shown between the two tcpdump sessions. This computer is one of 30-50 units (some with up to 250 UML instances) that we use to simulate thousands of computers. Originally I used bridging to connect uml-bridge to eth0 but with hundreds of ip addresses bridging onto a single network segment (with proxy arp enabled), arp caching became a problem. My solution was to route between the uml-bridge and eth0 (instead of adding eth0 to the bridge) and use zebra+rip for the route advertising. Functionally it works really well but inbound UDP packets in bursts are lost. TCP works just fine. Anyone have any ideas about those lost packets? ------------------------------------------------------- This SF.Net email is sponsored by: NEC IT Guy Games. How far can you shotput a projector? How fast can you ride your desk chair down the office luge track? If you want to score the big prize, get to know the little guy. Play to win an NEC 61" plasma display: http://www.necitguy.com/?r _______________________________________________ User-mode-linux-user mailing list User-mode-linux-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/user-mode-linux-user
