🚀 go-pugleaf

In article mea@nospambradnylion.com says...
> It does sound related and perhaps the reason ipv6 hasn't been a problem 
> is that hardly anyone uses it yet most all devices run some kind of ipv6 
> stack, maybe for testing and/or compatibility reasons.  Maybe people who 
> use MacBooks use a wifi product made by Apple that won't have this 
> problem.  I went back to check the October and November incidents and 
> they're both ipv6 as well.

Hashing this out here in writing was a good exercise for me because I 
think I now know the problem and for the alien archaeologists who one 
day dig up this newsgroup I'll post the solution.  The MacBooks or 
Powerbooks must be using ipv6 for some legitimate purposes on Apple's 
LAN products.  I'm getting a lot of MacBooks doing this now so Apple 
might be rolling this behaviour out more and more with recent OS 
releases.  Although IPads, IPhones, Ipods, etc. use ipv4 mdns, I haven't 
seen any of them us ipv6 mdns -- yet. 

That said, the problem was with a single vendor's AP/switch that caused 
a loop between the two of them.  Taking the second AP of that vendor off 
the network eliminated the loops when these multicasts occurred.  The 
topology of this network is very simple and there are no natural switch 
loops.  That vendor's switch was not handling the ipv6 multicast 
correctly and the problem with going el cheapo is that you get what you 
pay for.  I was able to narrow it down because I'm getting a lot of 
MacBooks or Powerbooks recently for some reason.

I could probably test this further but am not interested in debugging 
some off brand manufacturer's product that I got on sale for $12 at 
Microcenter (which is why I have more than one).  I think the MacBooks 
are a little abusive however and might have caused the hiccup.  In my 
recent trace with a successful session with my neighbor MacBook where 
she was able to do something and didn't cause a network crash, I counted 
112 mdns broadcasts with sometimes big 1400+ byte packets all within 
seconds upon entering the network.  All the APs now were able to handle 
this but for some reason I think this behavior contributed to trigging 
the loop in the off brand pair.  So Apple should share some blame in 
this -- IMHO.  There is no good reason to broadcast this tripe 112 
times.  After five what's the point?