There is tenant OBM and a MOM OBM to monitor the environment. Agents are managed by the tenant OBM which forwards the events to the MOM OBM. The issue is that despite MOM OBM is configured as a secondary manager for the agents, in some of the Nodes the performance perspective shows empty graphs while in tenant OBM the graphs are OK. From that MoM OBM jcodautil ping to the nodes work fine (telnet on 383 as well) when no graphs are shown. So, there is a secure connection in general.
Things that needs to be considered in such setup. To graph from PD, a node CI requires to have an Agent CI and the “Monitored by” set to “OM”. If the node CIs had something in the Monitored By attribute which didn’t include “OM” then the solution is to set “Operation Agent as Default Data Source” to true in Infra Settings. Also if the nodes on the MoM OBM have Primary DNS attribute populated with FQDN while in the local/tenant OMi the DNS is with short name one needs to change the DNS in RTSM for the MoM OBM to short name for the graphs to show. One also should clear the PD cache.
It was not working with DNS as FQDN because instance id in PD mapping UI for CI type Unix was set as display label but the CI's primary dns name property contained FQDN. So there was a mismatch. One can update PD instance id Mapping UI for the ci types nt and Unix to primary dns name and then do clear cache from PD UI and see if graphing works.
Instance Id mapping is used while graphing metrics from Agent and vPV.
For any ci that is monitored by agent/vPV, PD reads the value of instance id mapping to get the ci property using which it gets the classes list from the source(Agent/vPV).
In this setup, CI property Primary DNS name had -> FQDN.
Instance id mapping was -> display label/short name.
Agent returns the instance names with FQDN . Since there is a mismatch and PD is not listing any classes or gets data from Agent.
The problem in this case was that the Unix CIT is supposed to inherit the Primary DNS instance definition from the Computer type but here it was set to use display label on the Unix CIT which overwrites that. In that case PD uses the short name in the display label which is short name and does not match the FQDN logged on the OA from where the data is taken. On the other OBM it works with FDDN because there one do not have display label instance mapping on the Unix CIT, but uses the primary DNS mapping from Computer.
CI property defined in Instance ID mapping is used to filter the classes from Agent.
This should match with the instance in agent that we get from Coda API response of Agent.
If there is a mismatch, then PD fails to show the class list.
In your setup, instance id mapping was set to display label and coda API was returning instance name as FQDN . So there was a mismatch because of which classes were not getting listed in PD UI.
Micro Focus Software Support
The views expressed in my contributions are my own and do not necessarily reflect the views and strategy of Micro Focus.
If you find this or any post resolves your issue, please be sure to mark it as an accepted solution.