Kernel recommendations when NNMi is integrated with OBM
I have an enterprise deployment of NNMi integrated with an OBM. In reviewing the events, I have found that we are occasionally missing events, particularly closure events that would close the original NNMi event.
I had run into a problem with the UDP buffer in NNMi 9.2x integrated with OM when NNMi is running on a Windows system as the UDP buffer was too small and a series of about opsAgt events sent at once would overfill the buffer and get dropped (lost).
We are running on RHEL 7.7, and NNMi 10.50. Our buffers are set as per the guide but I am wondering given the integration if that puts additional strain on the buffer beyond what NNMi would normally need for a receive buffer. Particularly if we have too many traps coming in.
Recommended 10.50 receive buffer is 8Mb (8388608). I cannot seem to get my network team to turn off the syslog traps and the inteface linkup/down events so I think our load is notable at times and would consume a lot of udp resources on its own.
Any experience finding the appropriate value for UDP receive buffer on an NNMi with integrations to OBM?