[rbak-nsp] MPLS explicit-null drop on a VPLS service

[Christopher E. Brown]

Having an interesting issue with a provider supporting provider circuit
where I am the customer.

Simple ethernet service run from our MetroE network out through another
provider, connects a local MPLS node to a remote one.  All traffic is
flowing well *except* packets with an explicit-null label.


3rd party path is a redbak VPLS service that may have foundry/other
native switching on the far end.


Pretty much stumped on this one, everything but explicit-null labeled
packets are flowing just fine.


My thoughts were some form of auto-filter or service config eating
explicit-null labeled packets in the redbacks, or a bug in a label aware
device.

Maybe a ethernet LAG group or flow hash for multipath selection algo
that is looking at labels for part of the hash and disliking label 0/null.



Ideas?  We have no issues without our network, but we do not use the
same gear, looking for pointers/etc. to pass on to my peers on the other
side of this.


Thanks