v6ops at stefan-neufeind
Aug 17, 2012, 4:02 AM
Post #4 of 25
On 08/17/2012 12:57 PM, Jeroen Massar wrote:
Re: 1::1/128 + 2::2/128 - GRH Anomalies Delta (2012-08-17)
[In reply to]
> On 2012-08-17 12:41, Daniel Roesen wrote:
>> On Fri, Aug 17, 2012 at 11:57:51AM +0200, Jeroen Massar wrote:
>>> /128's.... and in non 2000::/3 space? Really!?
>>> ::/0 Default Route
>>> +1::1/128 Unallocated prefix
>>> +2::2/128 Unallocated prefix
>>> 64:ff9b::/96 Unallocated prefix
>>> 2000::/3 Default Route
>> The problem with GRH is that people tend to send unfiltered IBGP to it
>> instead of a properly filtered EBGP stream. Thus GRH sees all kind of
>> more-specific/internal junk that noone else will see, resulting in
>> "false positives".
> While that is true, 1::1 and 2::2 should never exist anywhere, they are
> outside of 2000::/3 and thus are not defined and thus should not be used
> and definitely never ever be routed on the Internet. Same for that silly
> 64:ff9b::/96 prefix.
>> To me, this report will only have a real meaning when all GRH peers are
>> only sending an EBGP feed which they would also send a downstream BGP
>> customer. Unfortunately, that's largely not the case so I stopped
>> looking at the report a looong time ago.
> That is true.
> Maybe it is time to request peers to send prefixes in a certain way indeed.
> We could also request people to send certain communities and instruct
> GRH to ignore prefixes with certain communities for some reports.
> Any proposals there?
As a first step imho contact peers and ask them to only send routes also
meant for others/downstreams, as Daniel said.