On Tue Sep 16, 2025 at 04:45:25PM +0200, Claudio Jeker wrote:
> On Tue, Sep 16, 2025 at 04:25:50PM +0200, Jan Klemkow wrote:
> > On Tue, Sep 16, 2025 at 03:25:21PM +0200, Rafael Sadowski wrote:
> > > On Tue Sep 16, 2025 at 03:18:28PM +0200, Jan Klemkow wrote:
> > > > On Tue, Sep 16, 2025 at 03:01:33PM +0200, Rafael Sadowski wrote:
> > > > > WireGuard shows severe performance degradation (95% bandwidth
> > > > > loss) on Intel 10Gb interfaces compared to direct connections,
> > > > > with significant packet loss patterns.
> > > > > 
> > > > > Performance Comparison:
> > > > > 
> > > > > ServerA (Chicago) - Intel 10Gb interface (ix0)
> > > > > ServerB (Atlanta) - Intel 10Gb interface (ix3)
> > > > > 
> > > > > - Direct connection (iperf): 66.8 Mbps
> > > > > - WireGuard tunnel (iperf): 3.3 Mbps
> > > > > - Performance loss: 95%
> > > > > 
> > > > > The physical Intel interface (ix3) shows 149426 output failures:
> > > > > 
> > > > > ix3     1500  <Link>      f8:f2:1e:3c:9c:09 195418012     0 144748154 
> > > > > 149426     0
> > > > > 
> > > > > suggesting hardware/driver level problems that worsen with
> > > > > WireGuard traffic processing?
> > > > > 
> > > > > Are there known compatibility issues between ix(4) driver and
> > > > > WireGuard packet processing?
> > > > > 
> > > > > Could the bridge configuration (veb0 + vport0) be contributing to
> > > > > the packet loss patterns?
> > > > 
> > > > Yes. You will lose that kind of performance over bridge(4) and veb(4)
> > > > because, they don't use segmentation offloading nor parallel processing
> > > > of packets.
> > > > 
> > > > The wg(4) device may have a similar missing performance features.
> > > > 
> > > > > Any guidance on debugging approaches or known workarounds would be
> > > > > greatly appreciated. I'm happy to provide additional data.
> > > > 
> > > > Could you provide a netstat -s stats, before and after you've done you
> > > > measurement?  So, we can see if there are also any error or drop counter
> > > > involved.
> > > 
> > > Before iperf:
> > > ...
> > > 
> > > After iperf via wg0
> > > ...
> > 
> > All counter are looking fine.
> > Nothing special for me.
> 
> I think the interesting stuff is missing. This feels like an MTU issue to
> me. I assume that the goal is to push traffic from veb1 via vport1 over
> wg(4). Now vport and veb have mtu 1500 but wg runs with mtu 1420.
> So either the packets get fragmented and then maybe dropped in transit or
> PMTU needs to kick in (which often enough fails). 

Thanks Claudio,

I tested with iperf --mss 1380 and 1420 with the same bad bandwidth.

Reply via email to