OLSR routing -help-

[rebel2234]

I have got OLSR running on 3 test boxes basically I have a center box that branches out to 2 different wireless routes with internet connections on the 2 ends via ethernet and I am connecting to the center box with an winxp machine to simulate a client.

I have been playing around all day with the test senario trying to get it to use one route over the other but cannot get it to use a particular route...

Here is my senario we have 2 differnet gateways to the internet one is a syncronous 10mbit and the other is 4X Dsl modems with a load balancer. What we wanna be able to do is use the 10mbit connection unless it fails then have the routes failover to the load balanced dsl gateways. Anyone else got a senario similar to ours and using OLSR? What do your OLSR configs look like? Can OLSR be used in this manner?

[tog]

Assuming you're using LQ (I use it, LinkQualityLevel 2) you'd use this inside your Interface "xxx0" {} section:


LinkQualityMult 192.168.0.1 0.40


See example config from wiki article (http://staros.tog.net/wiki/OLSR), the OLSR example config has that LinkQualityMult in there commented out as an example.

The LinkQualityMult line first argument is the direct peer system's IP and the second argument is the multiplier to assign to it. You need to assign a multiplier of 0.49 or below to de-prioritize that route.

As for determining whether the Internet connection is down or not, you'd use the dyngw plugin which inserts/removes a 0.0.0.0/0 Hna4 announcement at your border routers based on whether or not something returns pings.

I've never used the dyngw plugin myself but if I understand correctly, you do not put an Hna4 entry for 0.0.0.0/0 in manually at your border router, you let the dyngw plugin do that for you.

[rebel2234]

Thanks TOG! Works great! Yes Dyn_Gw plugin does add an hna of 0.0.0.0/0 for you. A couple of things here. First off on the WRAP's that I have enabled the dyn_gw plugin the processor usage is quite high 80% after disabling the plugin the processor usage settles at about 7%. Possibly a bug? I cant imagine a ping routine using up that much processor. After enabling the dyn_gw plugin my ethers lights stay pretty much steady (beacon says ICMP 12.54kb/s to my pinged gateway). I have tried setting the interval paramater to 50 500 and 5000 with very little change (beacon- ICMP 10.50kb/sec on 5000).

Also I see that you can specify your own broadcast address. The recommended is 255.255.255.255 but this would probably upset DHCP could one use a multicast address of say one similar to RIP like 224.0.0.9 or even 224.0.0.10 to get the job done? Anyone played around with the Broadcast setting?