Consitent intermittent latency/loss spikes with wireguard tunnel.

xparanoik

@heatmiser hahaha, yeah.. this wireguard thing is pretty sad. going to move my router to openbsd and use their wireguard which comes on base.

heatmiser

@xparanoik You mean Opnsense or actually doing this on a BSD? I'm building out a new Opnsense installation right now on a separate PC so I don't screw up my network.

I can't believe the debacle Pfsense has created here.

xparanoik

@heatmiser OpenBSD... I suggest you look into it, if you want a sane OS with really good documentation and a security focus. A few lines and you have a router, just not a GUI.

heatmiser

@xparanoik I'm quite familiar with Free and OpenBSD, used to host my servers with both, just never heard of anyone using it as a router appliance. Sounds cool though.

I'm gonna see how Opnsense works out for me. I seriously can't believe how bad Netgate blew this one.

Mullvad still has the same problem though no matter what we use, so we might want to refer back to this thread.....

xparanoik

@heatmiser Gotcha! Yeah, I just got a VM in the "cloud" and setup a wg tunnel to a VM at home, then I route traffic through that. I ended up disabling wireguard on pfSense and will build the new router on spare hardware so I can cut-over soon.

heatmiser

Anybody get any updates on this?

Still getting latency......

cmcdonald

@heatmiser You won't see any improvement until there is code updates. Patience

heatmiser

According to Mullvad this issue has been fixed.

https://mullvad.net/en/blog/2021/5/25/no-more-hiccups-or-maybe-just-one-more/

Anybody tested this?

xparanoik

@heatmiser This is good news. I will create a new tunnel and add it to my smokeping instance to see if the latency spikes are gone.

cmcdonald

@xparanoik Keep us posted. I'm very interested in hearing about these results.

I have not personally experienced the latency spikes across pfSense to pfSense tunnels

xparanoik

@theonemcdonald Will do. So far this issue has been specific to Mullvad (wg) tunnels. With their recent change sounds like it's fixed now, but I will get that monitoring configured and will post results here soon.