Using LMS 3.0.1 and some devices have never had inventory collected. If inventory collection is attempted it fails with a message "Device sensed but collection failed". The credentials have been checked for these devices and are ok. the devices in question are varying models connecting to either 3750s or 6509 switches. Any ideas about where I should be looking. This is a problem affecting only 10 out of 500 devices
I recently filed a new bug tracking a certain class of inventory collection problems, but there appears to be still one outstanding. Check your IC_Server.log. If you see errors saying an index for a table is not unique, then you are hitting at least one of these bugs, and you should open a TAC service request to get a patch, and to collect more debugging info.
The error occuring throughout the log file is: ASA Error -193: Primary key for table 'PhysicalPort' is not unique. Is this what you were referring to?
Yes. This problem will most likely be fixed by the patch I already have created (available through TAC for bug CSCsm97530).
I am getting the same error "device sensed but collection failed" on inventory. I can import the config and access the device and it passes the credentials tests. BUT I am not seeing the error about unique index in my ic_server.log file. Instead, I have the following in my log file:
[ Mon Mar 17 11:09:44 CST 2008 ],ERROR,[Thread-14],com.cisco.nm.rmeng.inventory.ics.core.CollectionController,667, Collection failed but sensed the device : 256
com.cisco.nm.xms.xdi.ags.system.CollectionFailedDeviceSensed: com.cisco.nm.xms.xdi.ags.system.CollectionFailedDeviceSensed: java.lang.NullPointerException
Would that patch help me at all or am I fighting something else?
rme is 4.0.6
both are 3750's
both run ver C3750-ADVIPSERVICESK 12.2.40 se
I have lots of other 3750's same code level no problem.
Do you have any pre-provisioned cluster switches configured on this 3750? That is, have you configured any cluster switches which don't yet exist? If so, remove them from the config, then try and collect inventory from this switch.
no, no clustering. These two switches do have cross-stack etherchannel back to the core switch, but so do other 3750's that don't have this problem....
Thank you in advance,
okay i've attached the shows per request.
I think I might see something here. Both systems in question seem to list 4 switches in the master election but both stacks only have 3 swtiches in play. It's been that way since before I started working here but I don't know what causes that. I suspected my predecessor took one out of the stack to deploy elsewhere but have no history to confirm this. I do know that I tried to force the stack to renumber so that it would be top to bottom 1, 2, and 3. All that did was jumble up the port configurations. If fixed it and, if you press the mode button, the stack is numbered 1, 2, and 3, in the order they are placed in the rack but...maybe the problem lies with the fact that it thinks somehow there are 4 switches in there?
Yes, this is what I said previously. You need to remove the switch which doesn't exist from the config. In this case:
no switch 4 provision ws-c3750g-48ts