IntroductionHow to reset (cleanup) discovery in NMS 7.x
We can use the same procedure as with previous versions, but it's a bit different because certain steps aren't applicable to NMS 7.0. The reason we say not to use it is that in large environments, it can really bog down the system. If the environment is small (a hundred or so robots), then it's fine, but thousands of robots would mean a large performance hit for a while.
The following procedure includes a WHERE clause in the DELETE statement. Again, this is only necessary if the environment is large and you want to target your deletions so everything doesn't get deleted at once. If the environment is small, just omit the WHERE clause and let it go.
ProcedureHow to reset (cleanup) discovery in NMS 7.x
Increase the discovery_server java memory min and max to at least 2048 and 4096 respectively.
Not having enough memory can cause inaccurate node/group nodes counts or produce unexpected discovery results in general.
Keep in mind that although you may want to delete a node from USM because its no longer needed, you might have leftover data that needs to be cleaned up in the QOS index table, S_QOS_DATA so check that first if you're performing a comprehensive cleanup of one or more nodes.
For every discovery_agent host:
?1. Deactivate the discovery_agent
?2. Deactivate the discovery_server
?3. Delete only the discovered computers you wish to remove from the database by
???? running the following queries:
????? delete from CM_COMPUTER_SYSTEM cs WHERE <some clause>;
????? Some clauses you can use include:
????? WHERE cs.NAME like '%somename%'
????? WHERE cs.IP = 'xxx.xxx.xxx.xxx'
????? WHERE cs.ORIGIN = 'someorigin'
WARNING - Deleting several hundred devices at a time can cause major performance issues.
IF you have a need to clean up all discovery and you can do it off hours after work or on the weekend you can 'delete from' without the WHERE clause for CM_COMPUTER_SYSTEM.
When you delete from CM_COMPUTER_SYSTEM, a cascade delete occurs so? CM_DEVICE and CM_NIMBUS_ROBOT is also wiped but you can still check if you want to be sure.
Also note that if devices have been deleted but are now reappearing in USM/those tables, that means discovery server is finding them in a niscache 'somewhere,' so the niscache has to be deleted and the target robot has to be restarted in all cases.
4. niscache deletion
To easily delete the niscache on robots/sources where you need to clean up nodes/Ips/robots/servers that either no longer exist, have no data, no monitoring etc., see attached probe package (niscache_clean_1.1.zip). You can drag and drop this package to delete the niscache contents and folder. Note that you MUST restart the robot/source system to reestablish the folder.
5. Restart nis_server to clean up the groups
At this time your discovery is empty. All your views should be empty. You may get transient 'failed to connect' messages, but that is expected.
Although you can adjust your discovery scopes at any time, now is a good time to adjust them if needed before restarting discovery.
Note that in CA UIM (NMS) version 8.0 you have to restart the ugs_server probe.
Once you are ready to restart discovery, follow these steps to do so:
??????? a. Activate discovery_server probe.
??????? b. Activate all discovery_agents.
??????? c. Wait for discovery to complete.
You can always run the Discovery Wizard and choose 'Run Discovery Now' and watch its progression then check the results.
Review the results in either UMP => Unified Service Manager, USM groups.
IMPORTANT: In large environments once the memory is increased and the procedure above is followed for a significant number of nodes or ALL nodes, it may take many hours for discovery to complete, e.g., 10 hours.
keywords: reset discovery UMP USM server agents cleanup clean data source targets groups nodes leftovers hubs robots systems devices device purge remove The system cannot find the path specified removal reappear reappears