DX Unified Infrastructure Management

Hi,

We have lots of confusion/discrepancies about Device Discovery in UIM Database

1. What is the difference between CM_DEVICE and CM_COMPUTER_SYSTEM table, what is the process to store any device in UIM DB after discovering it.

2. There are multiple entries of devices in UIM DB with same cs_id and different dev_id even for those devices also which we discovered only once. How it's happening , Is there any house keeping steps so that we can eliminate this duplicate.

3. Devices are being discovered in CA_UIM and suppose If any configuration of network device is being changed purposely for an example (Syslocation/IfAlias/IfOprstatus) then do we require to re-discover N/W device again or it'll update in UIM DB according to SNMP Collector Next poll cycle.

Kindly help us to understand above mentioned scenario as these are impacting our environment a lot.

1. The CM_COMPUTER_SYSTEM table contains reconciled assets which are displayed in USM. The CM_DEVICE table contains several perspective for the same Computer System. Let me picture that: if you install cdm, processes and the vmware probes for one of your system, you might end up with 3 records for that device in CM_DEVICE (because the device has 3 different perspectives, one for cdm, one for processes and one coming from the vmware probe) and based on reconciliation rules, it is all part of the same Computer System.

2. That might happen if you change the Device ID of a robot. There's a callback to do so. It also happens if you mess with the files on the robot and restart it or if you duplicate robots (with VM cloning already having a robot installed for example). You can then have several devices which are physically the same but are still producing several perspectives into the CM_DEVICE table, and based again on some rules, those devices might not be reconciled properly. All of the mechanisms involve discovery_server (and potentially qos_processor) and you can try to jungle with the discovery_server settings to mitigate this. You can also use some python rules in discovery_server to prevent this with proper naming (because, depending on the probe and the rules, several attributes might play a role: robot_name vs qos_source for a specific probe, vmware uses the vm_id and mac_address, etc.). You'll find several tips and explanation of settings and decision tree here: Device Correlation Configuration - CA Unified Infrastructure Management Probes - CA Technologies Documentation  and here: Device Correlation Troubleshooting - CA Unified Infrastructure Management Probes - CA Technologies Documentation you'll find a couple of troubleshooting steps.

3. For network devices, that depends on the kind of change. If that's a change in the IP address, you'll have to address the rediscovery, if that's a sysdescr, syslocation change, it should be taken automatically (but not for the profile name if you change the sysname). For Interface status, it will also be polled automatically but if you use rules based on this to apply templates, you'll have to wait for the template reevaluation or force it manually (in the case for instance where you exclude ifoper == DOWN interfaces from the polling, the change won't be reflected immediately in either way).

Hope this clarifies it a bit!

Christophe_Sperandio

Thanks for the response. 

1.I do understand in case of agent based monitoring through Nimsoft Robot. However It seems little strange.

* Do we have any specific reason to have 3 different entries in CM_DEVICE table if we have installed 3 probes on a robot machine that you are calling as    probe perspective of a device/Robot ?. 

* How does it varies in case of Network device discovery.

2. We are monitoring network devices through Discovery/SNMP Collector.We have multiple entries in CM_DEVICE table with different dev_id probably is will be due to network scope overlapping or can you please aware us from rest of the reasons ?

Our problem is , We need to create custom dashboard to show links based on their ifAlias value  .For creating this we need to join following tables CM_DEVICE, CM_CONFIGURATION_ITEM, CM_CONFIGURATION_ITEM_ATTR . We are getting lots of duplicate data in that dashboard while joining first two tables based on dev_id that is duplicate in CM_DEVICE for one device. We don't find any other relationship between tables.

* Can you please suggest us how we can handle such scenarios and maintain non duplicate records in CM_DEVICE table

* Which tables needs to refer in this case ?

3. I.P or host name change will require re-discovery anyways .  If following configuration ( IfAlias, IfOprStatus)  changes then they are not reflecting in neither in UIM DB nor in UIM Dashboards even after one week also. 

* is there any configuration in SNMP Collector which tells that what OID values needs to be polled during polling cycle so that we can include above    mentioned OIDs also if they are not there by default ?

Can you please educate us more on following. 

"For Interface status, it will also be polled automatically but if you use rules based on this to apply templates, you'll have to wait for the template reevaluation or force it manually (in the case for instance where you exclude ifoper == DOWN interfaces from the polling, the change won't be reflected immediately in either way)".

Please share any reference document if you have. 

Thanks a ton Chris

PFB too.

https://communities.ca.com/message/241847596?commentID=241847596#comment-241847596