<br><br><div class="gmail_quote">On Tue, Jul 24, 2012 at 1:49 PM, Teco Boot <span dir="ltr"><<a href="mailto:teco@inf-net.nl" target="_blank">teco@inf-net.nl</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
Op 24 jul. 2012, om 13:22 heeft Henning Rogge het volgende geschreven:<br>
<div class="im"><br>
> On 07/24/2012 01:15 PM, Teco Boot wrote:<br>
>>> Has been some time since I even looked into this code... the code<br>
>>> you removed were responsible for detecting a tunnel going down and<br>
>>> up again (because I thought it is necessary to set the route into<br>
>>> the tunnel again).<br>
>>><br>
>>> Its not really common, but could be potential annoying.<br>
>>><br>
>>> Can you test if what happens to the smart-gw if you take a tunnel<br>
>>> down and up again from the shell?<br>
>><br>
>> Manual interference with unmanaged tunnels is a request for trouble.<br>
>> If there is a side effect on tunnel_down (e.g. removed routes), that<br>
>> is how it is designed. Let's have a clean smart-gw.<br>
><br>
> At the moment there would be no way to restore the route for the tunnel. That is why the mechanism was inside.<br>
<br>
</div>Manual ifconfig xxx down ; ifconfig xxx up needs manual ip route add xxxx<br>
If tunnel can go down caused by unknown reason, and goes up by magic</blockquote><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">, then we may needs such.<br>
Better not depend on magic, and if so, lets task this magic with the complete job :-)<br>
(I could miss something here)<br></blockquote><div>imho if the smartgateway tunnel (which oslrd created) goes down, maybe olsrd should destroy/remove the tunnel. (-> doing this makes sure it can not go magically up again,..)</div>
<div><br></div><div>and afterwards (maybe with same delay as on intial olsrd start) look for a new smart-gateway and if found/choosen create a new tunnel.</div><div><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class="im"><br>
> Similar for mesh interfaces, if they go up again, OLSRd must restore the routes.<br>
</div>If iface goes down, all neighbor state should be cleaned up.<br>
After it goes up again, adjacency will follow and routes will be restored.<br></blockquote><div>afair this is exactly what currently happens, isn't it? </div></div>