Nortel 5000 Series Release Notes page 33

Change Request
Number
Q01720549
Q01721397
Q01722655
Q01723309
Q01723954
Q01728569
Copyright © 2008., Nortel Networks
.
Description
In an SMLT environment, all IST traffic (switched or routed) is blocked
from egress on SMLT / SLT ports that are currently in SMLT mode. This
is achieved through programming the PORT_TRUNK_EGRESS table
corresponding to the IST trunk appropriately. However, when the IST ports
and SMLT / SLT ports reside on the same 5695 device, the EGRESS_MASK
table needs to be programmed too. Currently, when an IST / SMLT port goes
up or down, the corresponding entry in the EGRESS_MASK table is cleared,
resulting in a loop. In the Regatta 5.0 code base, this situation is corrected by
re-executing the block IST call that reprograms the two tables correctly. This
call is made whenever an IST port goes up/down or an SMLT port comes up
and the trunk is already in SMLT mode.
To work around this problem, static routes must be used.
For the 10 G or XE port, the oversize counter does not increment.
For the 1 G or regular port, the oversize counter only increments if the packet
that it receives is between 1519 and 9216 bytes.
The vlacp port CLI command in Agent v5.1 no longer accepts multicast
MAC addresses for the funcmac-addr parameter.
•
When an ASCII configuration file uploaded from a switch running Agent:
v5.0.0 is downloaded to a switch running Agent: v5.1.0, an error,
Configuration script execution Failed, is generated if the
vlacp port ALL funcmac-addr command in the file is applied to a
multicast address.
• After you upgrade the agent from Agent: v5.0.0 to Agent: v5.1.0,
any multicast MAC address previously configured for the interface
funcmac-addr parameter needs to be manually modified to the 5.1
default value of 0.0.0 or a unicast address.
When you configure a DMLT as an IST or SMLT in a stack, Nortel
recommends that you have at least one DMLT link on the base unit. If you
have a DMLT link on the base and if the base goes out and temp-base takes
over, the traffic recovery process is much faster. Traffic loss is reduced.
Continuous observation of IPFIX data uses a lot of memory on the switch and
burdens the switch CPU to the point where it can't do other tasks. Therefore,
the switch checks the network traffic at each second beat, rather than looking
at every packet to maintain IPFIX records. When you use IPFIX, the traffic
volumes are an estimate rather than the actual measured flow volume.
Nortel Ethernet Routing Switch 5000 Series
Release Notes — Release 6.0
NN47200-400 04.01 Standard
6.0 12 December 2008
Known issues 33