Inventory collection jobs for Catalyst switches running 12.2(33)SXI code may fail with the "Device sensed; collection failed" message. This can occur in LMS 3.2 (RME 4.3) and LMS 4.0. The problem is that in 12.2(33)SXI code, the way the CISCO-FLASH-MIB data is fetched is done in a way to prevent overwhelming the supervisor. However, the downside of this is that SNMP operations take a bit longer. By default, RME/LMS has a two second SNMP timeout. This, as it turns out, is too small for properly collecting inventory from these switches. Not to worry, though; there is a workaround.
In LMS 3.2 (RME 4.3) go to RME > Devices > Device Management > RME Devices, and select the Catalyst 6500s exhibiting the problem.
![Screen shot 2011-02-12 at 7.24.00 PM.png](/legacyfs/online/legacy/5/9/9/10995-Screen%20shot%202011-02-12%20at%207.24.00%20PM.png)
Click Edit Device Attributes.
![Screen shot 2011-02-12 at 7.25.05 PM.png](/legacyfs/online/legacy/6/9/9/10996-Screen%20shot%202011-02-12%20at%207.25.05%20PM.png)
Click Inline Edit.
![Screen shot 2011-02-12 at 7.28.36 PM.png](/legacyfs/online/legacy/0/0/0/11000-Screen%20shot%202011-02-12%20at%207.28.36%20PM.png)
Select each device in the list, and set the SNMP Timeout to 10 seconds. Then click Modify. On the previous screen, click Apply to save the changes.
![Screen shot 2011-02-12 at 7.30.56 PM.png](/legacyfs/online/legacy/1/0/0/11001-Screen%20shot%202011-02-12%20at%207.30.56%20PM.png)
Then run a new Inventory Collection job. It should be successful.
NOTE: In LMS 4.0, this same interface exists but there is no link to it. Instead, point your browser directly to http://SERVER/rme/rmedevices.do where SERVER is the IP address or hostname of the LMS server (be sure to include the port as well if you are not using HTTPS).