Poor perfomance over OpenVPN
-
Hi everyone!
I have the next specs:
- FW: Super Micro XG-1537
- Fiber: 600Mbps symmetric
- VPN: OpenVPN
I'm the only connected to the VPN and the maximum throughput I get is 30Mbps when testing it via Iperf. No one else is using the fiber line.
The client fiber is 300Mbps which I'm connected.
These are my current settings:
![advance configuration.png]Very (/assets/uploads/files/1596797729924-advance-configuration.png)I've tried different Encryption Algorithm, buffer sizes, UPD fast IO, both transport protocols (TCP/UDP), enabled cryptographic hardware.
I'm quite desperate so any help would be greatly appreciated.
Thank you so much for your help in advance!
-
@jordiSL said in Poor perfomance over OpenVPN:
Super Micro XG-1537
How much are you getting on the same line without vpn?
Try running htop from cli while doing iperf.
And try not to run any iperf on pfopenvpn doesn't benefit from multicores, so if one core saturates during testing, that could be the issue.
-
You can also check this issue:
https://redmine.pfsense.org/issues/10311 -
@jordiSL said in Poor perfomance over OpenVPN:
FW: Super Micro XG-1537
You mean, like original Netgate hardware (XG-1537)?
@jordiSL "I get is 30Mbps"
Yes, it seems low...
(This gives me almost 10x higher speed (M11SDV-4C-LN4F), so your device also needs to know this speed)- interesting to read this:
https://docs.netgate.com/pfsense/en/latest/book/hardware/hardware-sizing-guidance.html
two things I'm thinking about now:
- loader.conf.local (flow control (FC), EEE, hw.igb.rx_process_limit="-1"
hw.igb.tx_process_limit="-1, etc.)
https://docs.netgate.com/pfsense/en/latest/hardware/tuning-and-troubleshooting-network-cards.html
https://calomel.org/freebsd_network_tuning.html- @jordiSL "The client fiber is 300Mbps which I'm connected."
incorrectly configured this side or incorrect measurement method...
BTW:
OpenVPN dslreports.com (on 500/200 - ISP):
and what about these?
ifconfig igb0 -rxcsum -rxcsum6 -txcsum -txcsum6 -lro -tso -vlanhwtso
(igb, IX, em, etc.)
- interesting to read this: