Troubleshooting Kafka duplication

Data duplication for downstream applications

If statistics collection for a telemetry type is set up in both NFM-P and NSP, the telemetry framework may receive the same information twice. This could lead to incorrect results generated by the telemetry clients consuming the duplicated data. To avoid this issue, you need to remove the NFM-P as a telemetry provider where necessary.

Perform one or more of the following:

  • Disable the collection of all telemetry by NFM-P: disable the MIB/MIB Entry policy, the accounting policy, or both.

  • If you have more than one client receiving information, you may prefer to use NFM-P (SNMP) for some telemetry types, and NSP (gNMI) for others. In that case, you can disable NFM-P as a telemetry provider for specific telemetry types.

    1. From the standby NFM-P site remove the mapping for each telemetry type where duplicate collection would occur.

      The NFM-P mapping files are found in /opt/nsp/nfmp/server/nms/config/statsDevices.  

    2. Restart the NFM-P server.

    3. Repeat the above for the alternate site.

    Note: The mapping files may be restored upon NFM-P upgrade, requiring this sequence to be performed again.

  • In NSP:

    Ensure that the classic NE is discovered with a unified discovery rule including the required mediation (gRPC and/or FTP respectively when collecting gNMI and/or accounting telemetry from the NE).  This ensures that NSP is the only telemetry provider, eliminating the possibility of duplicates.

    Note: If NSP Analytics is deployed ensure that only one of NFM-P or NSP is persisting the collected statistics.