Intro & Network Client Speed Problem



  • First post, so hello to all! Getting to know pfSense and loving it already. That said, I am having a problem I was hoping maybe someone could throw some ideas at.

    I have had my pfSense router setup and functioning now for about a month. My network is configured as such:

    • Cable internet connection, 100mb (Located in Germany)
    • pfSense box: Single core 2.4ghz Atom with 1Gb RAM. Server is running in an ESXi instance on an Octocore Atom 2.4ghz Supermicro mobo. HDD is solid state. 16GB total ECC Memory. 8 NIC's total on the physical box; 1 NIC dedicated to the WAN connection, the others are in a pool managed by ESXi.
    • Ubiquity Edgemax 48Port POE/500W switch
    • Ubiquity AP-AC-Pro Wireless AP. All Windows clients are using the AC 5Ghz radio.
    • Windows 2016 Active Directory (serves as DHCP and DNS)
    • DNS is setup as such: Windows AD servers have DNS installed (requirement) and point to the pfSense router for the Forwarder.  pfSense has the 2 DNS servers from my VPN provider on the General Settings tab and then I've enabled DNS Resolver in pfSense with DNS Query Forwarding enabled.
    • I have a VPN service that I use to get back to the USA for stuff like Netflix and Amazon Prime video. I use a second Interface to route that traffic to and only have my FireTV boxes on that interface, everytyhing else goes out the WAN to my Germany ISP routes.

    Now, the problem. I have 3 workstations on the network and all 3 use the 5Ghz wireless. 2 of the systems function just fine, I get descent transfer rates but the 3rd one I have a bottleneck. I have tested several ways and get similar results. For a quick gauge I usually use fast.com (Netflix speed tester) which I know is probably not accurate, but it gives me a quick idea. The site tells me on the 2 normal systems that I'm getting between 65 and 76Mbps but on the slow one I get about 2Mbps. Next I use the test that my ISP provides, again similar results; between 60-70Mbps on the normal systems but only 3Mbps on the slow one.

    So, then I tried iperf and I get even different results. On the 2 normal systems I get about 6mbit and on the slow one I get about 3mbit. I'm not clear WHY the drastic difference in the output form iperf verses the online testers (maybe I just don't understand iperf…it is my first time using it, so that's likely.)

    Bottom line, no matter what I use to test, one of my machines is throttled but the other ones are fine. This didn't start happening until just recently...like 3-4 days ago I'd estimate. It's possible the WiFi NIC (Alfa AWUS036AC) went bad, but it's only 2 months old so I have my doubts. I re-enabled the on-board NIC, which is only 802.11N and I did get about 6-8 mbit (which is better, and "normal" for that NIC and the reason I installed the Alfa), so this leads to to either the NIC is bad or pfSense has some sort of rule that blocked and is bottlenecking that MAC. Is this possible?

    Being relatively new to pfSense I may have messed something up and maybe just don't know where to look to solve this issue. I've looked in my firewall logs and don't see anything. I don't have any rules setup to block anything special to that laptop. I do run Suricata and the Snort rules, but I have disabled that service and ran the tests without seeing any difference.

    I'm at my wits end here. Anyone got any ideas?

    Really appreciate any help.
    Thanks,
    Randy



  • Good detailed description of your setup, thanks!

    Have you tried plugging in the slow host and running the speed test?  If the problem goes away, it is definitely related to the wireless in some way.



  • Good recommendation, will try. I suspect it will be faster, as I did get about 3-5mb faster when I swapped out the AC adapter for the on-board N adapter. This led me to believe that it had something to do with that NIC and that it was somehow tied to the MAC of that NIC. I was wondering if there was somewhere in pfSense that I could find that it's blocked or throttled by the MAC/IP.

    I'll grab a cable and test though…just to see.

    Thanks,
    Randy