About 30 1120s lost their nvram after a building power outage. They were running 12.2(15)XR.
Via show cdp neighbor detail on the Catalyst 4500 they show hostname "AP" with no ip address. A single 4500 can have some APs with configs and some APs without configs.
I configured all the APs in a conference room via the IPSU using the device's mac address and a 3550-PWR. I had to telnet into the 1120 and do a write mem before unpluging the 1120s to place into the ceiling.
If I try and run IPSU on the same L2 vlan I get no response. show mac-address-table on the 4500 indicates the correct mac is being learned on the port. I compared the 4500s trunk port configuration from before and after the power outage. If I hold in the 1120s reset button for factory default I can then configure the 1120 via MAC address and IPSU.
Any suggestions? I searched the bug toolkit with no hits.
Current options include going through the ceiling, removing the APs, reset to default and re-ip. This would take about 1 week of labor, and nothing is to keep this from happening again. We used Kiwi Cattools and Kiwi Syslog to record everything, the syslogs showed nothing of value. Cattools recorded a "write memory" had been sent to all devices on 7/15/2004, each time a config change was made via Cattools the checkbox for write memory was checked.
Since the 1120 doesn't have a console port, can TAC crack open the 1120 case and somehow do a post-analysis of the nvram contents? I'd like to know what happened without guessing the root problem.
