After enabling BGP traps on several devices, the following errors are observed in the VNM.OUT and the CPU utilization of the SpectroSERVER is high:
Apr 20 08:13:32 WARNING at CsHPSERequestSender.cc(575): SNMP send took 1511ms! Device IP: 10.xxx.xxx.xxx Request ID: 144562 Operation Type: 0 Community: #v3/P:SHA^xxxxxxxx/AES256^xxxxxxxx/smcs
SendRetCode: 0 usReturnCode: 0 Timeout: 3000 Device Port: 161 m_strCommunity: smcs
m_nVersion: 3 m_strContextName: m_nSecLevel: 3 m_nAuthProtocol: 2
m_strAuthPwd: care_nwk_smcs_authentication m_nPrivProtocol: 4 m_strPrivPwd: care_nwk_aes
Apr 20 08:13:32 WARNING at CsHPSERequestSender.cc(1707): timeout mutex lock took 1493ms!
If BGP is disabled on the devices, then the above messages are no longer observed and CPU utilization goes back to normal.
Release : 20.2
Component : Spectrum Core / SpectroSERVER
So as to be able to process the v3 BGP traps, SNMPv3 profiles need to be added in Spectrum and used to discover the associated devices.
After doing this, then re-enabling BGP on the devices, the traps can now be processed and CPU utilization should be normal.