pfSense NTP node (Or any heavy UDP) traffic tuning



  • I was tuning my pfSense for a Gigabit NTP node that runs behind it and I came up with this configuration to keep my pfsense stable.

    kern.ipc.nmbclusters="1000000"
    kern.ipc.nmbjumbop="524288"
    hw.ix.max_interrupt_rate="64000"
    net.isr.maxthreads="-1"
    net.isr.bindthreads="1"
    net.pf.source_nodes_hashsize="1048576"
    

    NOTE: hw.ix is specific to your NIC drivers, so you might have to tune this to your own NIC driver.


  • LAYER 8 Global Moderator

    How is this suppose to help anyone?

    You have not explained any of your reasoning behind any of your numbers. Nor explained what makes your traffic heavier than normal, etc.. I run a ntp server behind pfsense that is part of the ntp pool - does that make it heavy?

    If your going to post such info - you need to explain the reasoning behind each setting.. Or its just pointless noise.

    The person that understand what these settings do has zero reason to look for these.. So the only one that might find these and use them is someone that doesn't understand them - and you have provided zero insight to what they might do to solve some issue he/she is having..



  • Do you run your NTP server at the 1000Mb level?

    Anyone who is running a heavy traffic UDP server and is getting kernel panic crashes will come across this thread. They have every reason to have this info. This will help them and save them weeks of troubleshooting.

    If you don't tune your NIC drivers for heavy traffic you'll run into kernel panics. See thread: https://forum.netgate.com/topic/144075/ryzen-3-2200g-randomly-crashing/22

    And existing (Not as fine tuned) documentation: https://docs.netgate.com/pfsense/en/latest/hardware/tuning-and-troubleshooting-network-cards.html

    This just adds on to that existing NIC tuning doc for handling heavier traffic loads.


Log in to reply