[j-nsp] MC-LAG spanning-tree to be or not to be?

Sukhjit Hayre sukhjit.hayre at googlemail.com
Mon Apr 27 19:10:26 EDT 2015

hi list

I am confused on juniper terminology with regards to whether we really need
to run any spanning-tree variants when I have created mc-lag with mc-ae's
in family bridge mode....

guide here suggests not:


"Multichassis link aggregation groups (MC-LAGs) enable a client device to
form a logical LAG interface between two MC-LAG peers (for example, EX9200
switches). *An MC-LAG provides redundancy and load balancing between the
two MC-LAG peers, multihoming support, and a loop-free Layer 2 network
without running STP*."

latter steps 9,10 and 11 here suggest if we're dual homing into a non
spanning-tree device then block any bpdu's which I understand:


my understanding would be the following, block any stp bpdu's to any dual
homed devices whether they run spanning-tree or not (where not would be a
safety net just in case)

by default the mc-lag peer's set the DL & DR flags for man-addresses learnt

D -dynamic MAC
L -locally learned


D -dynamic MAC
R -Remote PE MAC

where one of the MC-LAG peers with the D & R flags will use the ICL-PL
(Inter-Chassis-Link Protection-Link) to ensure a L2 loop free path

would that be correct?

I also have the following two questions:

1.what decides the DL and DR flags? is it hash internal algorithm?
2.what are the re-convergence timers i.e hold/flush timers should the
mc-lag peer with the D&L flags have his link broken so that the mc-lag peer
with the D&R link changes it to D&L

thanks in advance for any replies

More information about the juniper-nsp mailing list