Hi, I have a 3845 with IOS 12.4(11)T1 (needs this IOS for CME).
I am attempting to do simple redundancy on the gigabit ethernet by setting G0/1 as a backup interface to G0/0 using "backup interface G0/1" on G0/0.
All is working fine apart from one problem. Everytime I reboot the router, both internal gigabit ethernet interfaces have the "no keepalives" command assigned to them.
This command is not in the startup config and is also not the default. This occurs whether the ethernet ports are connected to a switch or not.
This stops the redudancy working as the router does not detect the gigabit ethernet interface going down.
I am afraid you are hitting this IOS bug.
CSCsg68683 Bug Details
Headline c3800: GigabitEthernet interface keepalive settings change at boot-time
Feature 3800 Component Components Duplicate of
Severity 3 Severity help Status Resolved Status help
First Found-in Version 12.4(10.5), 12.4(10.5)T, 12.4(08b), 12.4(10a), 12.4(12.2)T, 12.4(11.06b) All affected versions First Fixed-in Version 12.4(9)T03, 12.4(12a), 12.4(12.15), 12.4(12.15)T, 12.4(07e) Version help
The keepalive N command under a GigabitEthernet port changes its setting at
router boot-time. For the following startup-config:
ip address dhcp
no ip address
we see the following happen at boot-time:
(1) GigE0/0 with keepalive 15 NVGENs to not having the command show at all --
it thinks the keepalive has been set to defaults.
(2) GigE0/1 with no explicit keepalive N command NVGENs no keepalive.
After boot-time it is found that the following happens:
(3) If I assert keepalive N on either interface then the desired command does
NVGEN, but will be affected by (1) at the next reboot if the current config
is saved to NVRAM.
(4) If I set default keepalive, then there is no keepalive N command NVGENed
-- it thinks the keepalive has been set to defaults. It is affected by (2)
at the next reboot if the current config is saved to NVRAM.
This behaviour WILL BE observed on Cisco 3825 and 3845 routers which are
installed with an IOS specified by or implied by the "First Fixed-in" field of
bug ID CSCek46943 "700-800 msec ping delay occurs for Gig-eth int of c3800".
This is a sufficient condition for the problem symptom to occur; no special
configuration is necessary. This problem is not observed on a Cisco 2821 or
2851 installed with the same version of IOS.
There is no known workaround other than to use an unaffected IOS release.