Traffic data wrong
when performing a throughput test, I noticed that PFsense does not calculates the traffic as should. When performing an outbound performance test with 57 mbps up and 247 mbps down, it looks like PFsense only can measure 13 mbps up and 12 mbps down. I Also checked the other interfaces, but I am sure the outbound traffic goes via WAN_PRIMARY.
What could be the reason that the traffic is not measured right?
I know of an outstanding issue where the traffic double counts data on a vlan'd interface, but this one seems different.
Something seems a bit off to me. They graph on Tamosoft doesn't even remotely match the shape on PFSense. Not only that, but the ratios between up and down are off. Are you sure the data is going through PFSense? How is your network setup?
How many WANs do you have? How is the traffic routed?
Like Harvy said your traffic is just way off what it seems it should be. Seems like something is fundamentally wrong.
thanks for the feedback!
I have 3 WAN interfaces:
and I haven 3 LAN interfaces:
Inbound I have added soms NAT rules for routing traffic on certain WAN interfaces to some specific lan interfaces/lan servers.
Outbound traffic all goes via WAN_primary (because I did not manage yet to route outbound traffic trough another interface yet, see https://forum.pfsense.org/index.php?topic=84807.0).
I checked graphs of all other interfaces. And I see increased traffic on WAN_primary and LAN_5_DMZ (where the Tamosoft client is located) when I run the test. When I stop the test, the increased traffic stops. I don't see increased traffic on the other interfaces. So it looks like traffic goes trough the right interfaces.
Also, at the remote site where the Tamosoft server is running, I see that I am connecting via the IP that is assigned to WAN_primary.
The traffic graphs could be averaging if the test is very 'pulsey'. You could try using a command line utility to measure the throughput. Maybe iftop:
You are right! I just tested using speedtest.net, and now I see the 800Mbps peaks without issue.