On Tue, Jan 03, 2006 at 11:28:47AM +0100, <Alpt>: ~> On Sun, Dec 25, 2005 at 11:29:21AM +0100, <Alpt>: ~> ~> The multipath code creates new cached routes. Since ~> ~> after connect the socket is "connected", i.e. saddr and ~> ~> daddr are known and they are always provided when resolving ~> ~> route ~> ~> ~> ~> So, the connected socket should not experience any outage ~> ~> when the route is resolved after cache entry expiration ~> ~> assuming the routing rules do not change. ~> ~> (That is snip of the Julian Anastasov's reply in a lartc thread: ~> ~> http://mailman.ds9a.nl/pipermail/lartc/2002q2/003780.html) ~> ~> ~> ~> Does it mean that it should work? ~> ~> No, it doesn't. Any ideas? ~> ~> PS: this thread was: ~> http://marc.theaimsgroup.com/?l=linux-net&m=113550638110682&w=2 Yay, there is a solution, but we cannot use it: The IP of gw A is 10.157.108.4 The IP of gw B is 10.29.212.6 The IP of C is 10.157.108.213 They are in an ad-hoc network, A can see C but not B, and B can see C. The steps are: - the node C creates for each gw an IPIP tunnel to A and a second to B tunl0: ip/ip remote any local any ttl inherit nopmtudisc tunl1: ip/ip remote any local 10.157.108.214 ttl inherit - the node C assigna a different IP to each tunnel ifconfig tunl0 10.157.108.213 ifconfig tunl1 10.157.108.214 At this point we have two different interfaces and we can use the LARTC method: http://www.lartc.org/howto/lartc.rpdb.multiple-links.html an we have: pc1:~/src# ip rule 0: from all lookup local 32764: from 10.157.108.213 lookup 201 32765: from 10.157.108.214 lookup 202 32766: from all lookup main 32767: from all lookup default pc1:~/src# ip route show table 201 10.0.0.0/8 dev tunl0 scope link src 10.157.108.213 default via 10.157.108.4 dev tunl0 pc1:~/src# ip route show table 202 10.0.0.0/8 dev tunl1 scope link src 10.157.108.214 default via 10.29.212.6 dev tunl1 The default route in the main table is: default nexthop via 10.157.108.4 dev tunl0 weight 1 nexthop via 10.29.212.6 dev tunl1 weight 1 All the rp_filter controls are disabled and ip_forwarding is enabled. It works, in fact, each time the routing cache has to be filled, a random route is chosen from the multipath route and all the packets which have a specific source go out from the same route. Now the problem is that a node has to have only one IP! If we assign other IPs to the tunnel interfaces conflicts will arise because those IPs will be reachable. I've tried to use `iif' in the routing rules but it doesn't work if the same IP is assigned to all the tunnels. Is there a nice solution? Best regards -- :wq! "I don't know nothing" The One Who reached the Thinking Matter '.' [ Alpt --- Freaknet Medialab ] [ GPG Key ID 441CF0EE ] [ Key fingerprint = 8B02 26E8 831A 7BB9 81A9 5277 BFF8 037E 441C F0EE ] - : send the line "unsubscribe linux-net" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
