Hi Kevin Am 12.05.2014 09:34, schrieb Kevin Boatswain: > > Thank you for the response I wasn't sure if anyone would be up at this time > to help (2:30 am central time us where i am at). That's when it's morning in other regions.
> > I ended up trying these settings in the /boot/loader.conf.local > > kern.ipc.nmbclusters="131072" > hw.igb.num_queues=4 Depending on available memory look at the mbuf usage in pfSense UI if it exhausts it under load inclrease it, otherwise leave as is. However i.e. for 10GE adapters Intel recommends larger nmbcluster size.* > I however am not sure if these settings are appropiate for my setup or not. There is no patented recipe for this, however the values in the pfSense Wiki correlate with other known good values shared n the FreeBSD universe (i.e. FreeNAS). So they must be pretty much proven / OK. > My box does currently have 4 igb nics (intel i354 x 4) and also currently > has 8 cores (c2756) . > > Does this mean I should try hw.igb.num_queues=8 instead of > hw.igb.num_queues=4 ? Also here it depends: Test and see if you are fine with the results, otherwise tune. It depends on the workload you throw at the box and also how many other services you gonna run on it. > I am not familiar with these settings just trying to figure out what > settings I should apply for stability and out of the box performance. I wasn't too and neither am familiar now. The base pfSense settings are often chosen on the basis of less-powerful boxes ** (to not exhaust limited resources) that's why you have to tune things a bit. -- Mathieu * http://downloadmirror.intel.com/14688/eng/README.txt ** Something like chooseable pre-tunings for slower or or larger systems would be interesting. :-) _______________________________________________ List mailing list List@lists.pfsense.org https://lists.pfsense.org/mailman/listinfo/list
