Upload issues in Windows 10 on PF v2.6.0
-
Has anyone noticed any issues with their upload speed in Windows 10 diminish to complete crap after upgrading to 2.6.0? To spare length:
Key Details: Windows 10 Client PC, fully patched, with Windows NIC driver, OEM driver, and various OEM driver versions, all exhibit the same behavior in upload speed with PF in the picture. Note: I can connect my Win10 Client PC to the modem and get expected upload speeds but when PF is introduced, upload goes to crap. (Expected is ~50Mbps, reality is between 6-15Mbps with PF on Win10 PC).There are no running services such as Snort, Wireguard HAProxy, Bonds, etc in place while testing (Although other systems show no signs of issues while these services are running)
WiFi 6 AP where all wireless client devices pull the expected upload speeds.
Ubuntu server that pulls expected upload speeds.
There are several similar reports but seemingly mostly reference Hyper-V. I do have Virtual Box installed but all tunnel adapters have been removed/disabled including features on the NIC.
In following the related workaround here - https://www.reddit.com/r/PFSENSE/comments/sukl9z/comment/i2zczjx/?utm_source=share&utm_medium=web2x&context=3 I did notice improvement in upload, peaking around 20Mbps but taking a while to get there where other systems pull 20-25Mbps almost instantly.
The second most successful resolution to other reporting this issue is to roll back to 2.5.2 which is a painstaking option in my case since I don't have a way to restore config backward compatible. It's worth mentioning, I too didn't have this issue until 2.6.0 which further confirms this has to be an issue in PF somewhere.
I'm running PF on a Watchguard XTM 5 series, fully up to date
.
Edit: Tested between another router and all is fine. Just purchased another NIC for further testing. -
That RSC issue only affects pfSense running as a VM in Hyper-V.
So the same client connected via wifi gets the expected upload speed? And that still goes through pfSense?
Restrictions if that sort of order are usually either unintentional traffic shaping or a speed-duplex mismatch in a link somewhere.
Steve
