Re: Point-to-Point interfaces regressions

From: Julian Elischer <julian_at_elischer.org>
Date: Fri, 25 Sep 2009 09:29:19 -0700
John Hay wrote:
> On Fri, Sep 25, 2009 at 12:27:04AM -0700, Li, Qing wrote:
>>> Li, Qing wrote:
>>>>> Me and many other people running net/mpd handling thousands of PtP
>>>>> interfaces sharing local addresses with each other and with some
>>>>> Ethernet interface. This change makes such setup inoperable, as mpd
>>>>> will constantly receive errors while trying to set addresses and
>>>>> drop connections.
>>>>   I can revert to the old behavior. I asked for feedback and
>> received
>>>>   none at the time ...
>>> I would like to see it working, but working correctly. If it is not
>>> possible to implement it correctly - then reverted, or at least
>>> disabled
>>> by default.
>>>
>>   Asking for compatibility is reasonable but the existing packet 
>>   leaking behavior is incorrect.  
>>
>>   After reviewing the previous thread on this topic back in July,
>>   I see that Henri Hennebert requested this feature for IPv6.
>>
>>   I intend to introduce a sysctl variable to control whether loopback
>>   route should be installed for the local end point. By default
>>   the old behavior maintained.
> 
> What about only adding the route if it does not exist yet? That should
> handle the common case of reusing your ip address on the ethernet
> interface on the local side of point-to-point links.

the hard part is removing it if it is the last interface that uses it.

> 
> John
Received on Fri Sep 25 2009 - 14:29:17 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:56 UTC