Ipv6 vs V2H124 : multicast strikes the stack

  • 0
  • 1
  • Problem
  • Updated 3 years ago
  • Solved
Hi community, I'm in trouble with the V2H124 stack.
We are renewing a lot of PC with Seven OS and for three time we encountered this situation :

when a new pc is connected to a stack of V2H124 the stack became crazy and instable.

It seems like a network loop caused by the IPV6 multicast (probably, of course ... it's a supposition according with our analysis and observations).

Anyone have suggestion (except "replace that old devices" :-)  ) or confirmation about this issue ?

Someone can confirm are V2H124 IPv6 compliant?

Thanks to all those who can help me, Michele.
Photo of SC-Lantech

SC-Lantech

  • 124 Points 100 badge 2x thumb
  • surprised

Posted 3 years ago

  • 0
  • 1
Photo of Mike D

Mike D, Alum

  • 3,852 Points 3k badge 2x thumb
Hello Michele,

No compliance doc or cert exists for the V2H.   
IPv6 introduces some new protocols but still has to run on ethernet.  Any layer 2 ethernet switch will be able to handle IPv6 forwarding.   
ARP goes away in v6 but plenty of multicast is present.  IPv6 multicast packets get the same treatment IPv4 multicast receives; forward out all ports except the ingress port. 
 
If I'm on target, you're not describing packet replication observed in a wireshark trace.  I gather you've determined it's not a loop - but the multicast aspect of the problem and the crushing impact on the network is that of a loop. 
Most of us are familiar with the confusion and chaos L2 loops bring to the party. 

We hear similar symptoms described from time to time in support.  It seems to happen to all switch types. The behavior is destructive to network switches and the L2 network.  There appears to be more than one flavor of the problem described below (link: nic blasts ipv6 mld queries).  The good news - if this is your problem there's a solution: driver updates.
The following link (from a quick search 'ipv6 sleeping nic') is not an official bug report, it's a journal/blog entry from a battered network admin.
 There are others like it:
http://packetpushers.net/good-nics-bad-things-blast-ipv6-multicast-listener-discovery-queries/
 
Often this starts as an annoying slowdown eventually getting very painful.
The behavior is ddos-ish since many end stations will exhibit the condition at once.  OS upgrade rollout establishes the means -  then everyone in the physics dept goes home at the same time.  Many end stations will power down at around the same time. Silliness ensues.

My official recommendation:   Order 75 or 80  X460-G2's.  :)


Let us know how things are going Michele.  If this driver problem isn't a fit, bump the thread again. There's sure to be someone in the community with experience that will be able to help.

Regards,
Mike
Photo of Curtis Parish

Curtis Parish

  • 3,332 Points 3k badge 2x thumb
Don't feel too bad about your V2H.  We had some new Dell computers with old network drivers bring  down S-chassis switches.       When it comes to storms I  say that 10Gb just gets you in trouble 10 times faster than 1Gb.   I guess compared to the V2H a hundred times.