Datasheet
295
Release Notes for Cisco IOS Release 12.1E on the Catalyst 6500 and Cisco 7600 Supervisor Engine and MSFC
OL-2310-11
Caveats
• When data-link switching (DLSw) Ethernet redundancy is configured, the switch might reload with
a bus error if circuits are established while peer connections are torn down. This problem is resolved
in Release 12.1(8b)E6. (CSCdt82241)
• After entering shutdown and no shutdown commands on an interface running fast switching, some
of the route cache entries for directly connected hosts are not created correctly, which causes
network connectivity issues. This problem is resolved in Release 12.1(8b)E6. (CSCdv43186)
• Suboptimal routes might be installed into the routing table if more then the configured number of
equal cost paths exist. This problem is resolved in Release 12.1(8b)E6. (CSCdu41228)
Resolved Caveats in Release 12.1(8a)E5
• It is possible to send an ARP packet to a local Ethernet device running on specific versions of
Cisco IOS, which could cause that interface to stop sending and receiving traffic. ARP packets
received by the router for the router's own interface address but a different MAC address will
overwrite the router's MAC address in the ARP table with the one from the received ARP packet.
This was demonstrated to attendees of the Black Hat conference and should be considered to be
public knowledge. This attack is only successful against devices on the Ethernet segment local to
the attacker or attacking host.
This vulnerability is documented in Cisco Bug ID CSCdu81936, and a workaround is available.
The workaround for this vulnerability is to enter the router interface MAC address into the ARP
table with a configuration entry, sometimes known as “hard coding” the ARP table entry.
The syntax for this command for routers and switches running IOS is as follows:
arp ip_address hardware_address type [alias]
The syntax for this command for switches running Catalyst software is as follows:
set arp [dynamic | permanent | static] ip_address hardware_address
The caveat to this workaround is identified with defect CSCdv04366, which will clear all manually
entered MAC addresses from the ARP table, when they are the same as the interface MAC address,
when the command clear arp is issued on the router.
This problem is resolved in Release 12.1(8a)E5. (CSCdu81936)
• Some policy-based routing configurations that divert traffic to external devices for additional
processing (for example, to a firewall) might incorrectly divert the traffic again after it has been
processed by the external device. This problem is resolved in Release 12.1(8a)E5. (CSCds60420)
• An MSFC2 might reload if you remove a default static route that was recursively learned from
multiple paths. This problem is resolved in Release 12.1(8a)E5. (CSCdv46440)
• In an Anycast multicasting environment, the convergence time is too long when the link to one
rendezvous point (RP) goes down and the receiver’s designated router (DR) needs to send join
messages to another RP. This problem is resolved in Release 12.1(8a)E5. (CSCdv27799)
• When redistributing RIPv2 tagged routes to another routing protocol, route filtering fails in route
maps configured with the match tag command. This problem is resolved in Release 12.1(8a)E5.
(CSCdt71063)
• When heavy traffic causes significant TCP latency, closing an established connection might cause
memory leakage of the TCB, which you can display by entering the show tcp brief or show tcp
brief all command. The leaked TCB will be in the TIMEWAIT state. This problem is resolved in
Release 12.1(8a)E5. (CSCdu79634)