[Olsr-dev] some thinking about the constant metric

[Henning Rogge]

Hi,

I just pushed a new version of the OONF master branch. With this you
should be able to use the unspecified address ( "-" ) as a neighbor
value in the constant_metric plugin.

e.g.

[constant_metric=eth0]
         link     -     100



This should set the metric value for all neighbors on eth0.

Henning

On Mon, Oct 5, 2015 at 10:59 AM, work_xl <(spam-protected)> wrote:
> Thank you for your reply.
> Of course  I am willing to test the 'any' option for neighbor IP because I
> think I may use it in my own work.But I am really busy working these days so
> I can't guarantee the time.
> After all ,I will try to do the testing .
>
>
> ------------------ 原始邮件 ------------------
> 发件人: "Henning Rogge";<(spam-protected)>;
> 发送时间: 2015年9月30日(星期三) 晚上8:16
> 收件人: "work_xl"<(spam-protected)>;
> 抄送: "olsr-dev"<(spam-protected)>;
> 主题: Re: [Olsr-dev] some thinking about the constant metric
>
> On Tue, Sep 29, 2015 at 5:43 PM, work_xl <(spam-protected)> wrote:
>> Hi,
>> I am now reading the constant metric plugin code and have got a question
>> about the way to use it.A constant metric is bound to a link which is
>> indexed with a neighbour address and an interface name.The doc said the
>> we
>> can use 'any' to  as a wildcard(I didn't check it carefully in the source
>> code),but it seems that the neighbor address must not be 'any'.
>> Here is my thought, in many adhoc networks,we don't know the neighbors'
>> addresses before we "neighbor" them . The olsrd2 would be easier to use if
>> we could just use 'any' to map all the neighbors which is linked through
>> one
>> interface.In heterogeneous adhoc networks, a router often links different
>> radios through different interfaces.So interfaces have their own metrics
>> instead of globe metric is understandable.Is there any reason we don't use
>> 'any' neighbor address?Or Do I misunderstand the source code ?
>
> I have looked through the code... at the moment the "any" option is
> just possible for the interface name, not for the neighbor IP. I will
> put this on my TODO list because it looks like an useful feature.
>
> If I get the feature running, would you be willing to test it?
>
> Henning Rogge