DComTalk.com

Having some problems setting up a stack of 5510's in a redundant Layer
3 configuration.

In my test lab (2) 5510's are stacked to make up the core. Both access
layer switches (also L3 5510's) have a Multi-Link Trunk consisting of
two interfaces back to the core. The MLT's are configured to act as
Point-to-Point links by using a unique VLAN between the core and each
access layer switch. Other VLANs exist on each access layer switch but
do not extend their broadcast domains into the core. Static routes are
configured on each device for each subnet (5510's do not yet support
OSPF.) When all devices are operational everything works including
failover on the MLT's. Pull any ethernet cable and the network
converges just fine. However, if the "Master" of the stack loses power
it no longer routes IP. L2 connectivity works just fine but IP routing
dies.

If I log into the secondary switch in the stack, disjoin it from the
stack, issue the following command - "no ip blocking", reboot and issue
the "ip routing" command I can make routing work by pointing the
client's gateway to the new stand along switch IP (NOT the stack IP.)

Any suggestions? Logic (and Nortel documentation) tells me that when
the master fails another device in the stack will assume the role and
the network should converge.

How did you configure the slave switch? Individually and join,or join
and let the slave be configured from the master? What about firmware&sw
revisions?

I tired to configure the stack both by the dipswitch in back and by the
"UI" button on the front panel. The dipswitch has been much more
consistent though neither way of setting the master allowed the other
switch to take over when the master failed. When I configured it with
the UI button sometimes the stack was never recognized. The "Base"
light would blink and each switch only used it's own IP, not the stack
IP.

BOSS 4.1 SSH
HW:03 FW:1.0.0.19 SW:v4.1.0.05