Hi - Curious if anyone encountered the followed. When running 4.8.4; speed tests with Surfshark VPN tunnel ENABLED on the GL-BE9300 router and VPN enabled on my PC (Note: my PC IP is not routed through the GL-BE9300 VPN Tunnel) meets 1gbps connection speed as expected. However however when I test the same with 4.9.0 installed, I lose about 400 mbps on the same speed test. When I disable either the VPN on the router OR the VPN on the computer, it goes back up to 1gps. Any thoughts on what’s causing this would be greatly appreciated!
Hi,
Can you try to add this at the end of /etc/hotplug.d/iface/99-ecm:
# 4.9.0 workaround: force ECM re-enable after sync to restore LAN->WAN acceleration
# WireGuard client flows will use software path but LAN->WAN will be hardware accelerated
[ -e /sys/kernel/debug/ecm/gl_front_end_ipv4_stop ] && \
echo 0 > /sys/kernel/debug/ecm/gl_front_end_ipv4_stop
I had the same issue with wireguard server, I reported it, maybe the same workaround works for you too...
Thanks Friend. I’ll give it a try in a little while. Had to roll the firmware back.
I have the same problem with Windscribe Wireguard. When running on my BE3600, the speed is too inconsistent to stream any video. I tried the app on my laptop and Android streamer without success. It seems that the VPN connection loses a lot bandwidth sometimes 90%. I’m in an RV and my current location has a slower connection. It’s fast enough for streaming with no VPN, but when I turn on the VPN, it constantly buffers. After trying everything else for a week, I reverted the BE3600 to 4.8.3 firmware and the loss of bandwidth due to the VPN is back to a normal 10-25% percent. I’ll wait until this has been addressed before I try to update again. Is it the processing power of this device or something in the new multi-tunnel VPN subsystem?
When I bought the router, i looked at the specs and at the time was convinced it was a strong processor, certainly enough to handle a VPN. I still have moves foward with the new firmware update as I don’t feel like “experimenting” at the moment. Time is also a factor. I’ll repost if I decide to mess around with it and have success.
Hi,
Thank you for providing the detailed comparison between firmware versions 4.8.3 and 4.9.0.
We performed a local comparison using the same WireGuard location and the same speed-test server. On firmware 4.9.0, the VPN throughput was slightly lower than on 4.8.3, but we did not reproduce the severe or inconsistent bandwidth loss you described.
Since VPN server load and routing conditions may vary between locations, could you please let us know which Windscribe location you were using and, if available, share the approximate download speed before enabling the VPN and after enabling it on both firmware versions?
Please also let us know whether the speed remains consistently low throughout the VPN connection, or whether it fluctuates significantly between repeated tests.
This information will help us determine whether the behavior is related to a specific VPN location or to the firmware version.
Thank you for your patience and assistance.
1 Like
I was originally using a static IP server located in Dallas, TX. I am located in northernwestern Washington State (near Port Townsend). I tried to switch to a couple of servers in Seattle (Cobain and Hendrix) without any improvement. I was using 5G home internet service. I then reactivated my Starlink account and was getting regularly greater than 100 Mbps speed tests. Windscribe nodes in Seattle and the static IP in Dallas varied from slightly under 100Mbs to less than 15 Mbps.
Since there are so many moving parts, I cannot swear that it was 4.9.0. Since I rolled back to 4.8.3, I have had almost no buffering and 95 Mbps (just tested) on the Dallas server.