We had a similar problem here, and after calling support to get help,
what the problem was eventualy determined to be was most likely the
result of a cut-and-paste operation between the IP Map and a
Collection submap. Apparently both Collections and the IP map
"nodes" are maintained in the database as objects, but they are
not of a compatible format, so a cut and paste between the two will
cause irrecoverable DB corruption (i.e restore from tape). There were a few
commands that support had me try to clear up
the bad Collections in the DB, but they failed.

Good luck,

Scott

>
> We migrated from NV Version 4 to V5 and everything worked out fine. But
> after the first reboot two daemons did not start: nvcold and C5d.
> If we start nvcold by using "ovstart -d nvcold" we get the following
> messages:
>
> ovstart: connected to ovspmd on attempt 1
> ovstart: sent START request to ovspmd
>  object manager name: nvcold
>  behavior:            OVs_WELL_BEHAVED
>  state:               NOT_RUNNING
>  PID:                 83850
>  last message:
>  exit status:         exit(1)
> ovstart: EOF from ovspmd, exiting
>
>
> If we try "/usr/OV/bin/nvcold -B" to start the daemon we see this
> message:
>
> ERROR: nvcold: Unable to add collection to list - is a duplicate name
> (file nl_main.C, line 702)
>
> Any ideas on this?
>
> TIA Richard Fitzinger
>


-----------------------------------------------------------
Scott Wilson                         Email: swilson AT rpm DOT com
Network Management Consultant        Pager: 800-506-7348

                     RPM Consulting, Inc.
                 7130 Minstrel Way, Suite 230
                     Columbia, MD 21045

Corp. office: 410-309-6000       Corp. fax:    410-309-6070
-----------------------------------------------------------