Date: Tue, 7 Feb 2017 01:21:41 -0500 From: Vincent Olivier <vincent@up4.com> To: FreeBSD virtualization <freebsd-virtualization@freebsd.org> Subject: Re: tap on lagg ? Message-ID: <CAD7158B-5481-4323-BBE7-3C9E22308180@up4.com> In-Reply-To: <CAJpsHY5b-oqvyXgiR9p-74=XeKB5-VmHs%2BkyoA2E=wOo8GuGWA@mail.gmail.com> References: <11A193E5-555F-4733-B192-49A5FEDCFDEA@up4.com> <66637AFA-D092-4EBB-B998-1BB2B2EE2CB4@up4.com> <1207dbb0-ec86-34fd-9a74-68d70b3b7892@osfux.nl> <CAJpsHY5b-oqvyXgiR9p-74=XeKB5-VmHs%2BkyoA2E=wOo8GuGWA@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
>> I have multiple machines on which 2 or more nics make up a LACP lagg >> with vlans on it. Those vlan interfaces are in bridges together with the >> tap interfaces that are in use by bhyve vms. >> >> Works as long as I "up" the nics in a specific fasion ( " -tso4 -lro >> -vlanhwtag " ) . This works on 10.3 and 11.0 as far as I'm aware and I >> have never experienced problems with it (Intel / AMD / em driver / bce >> driver, re driver , all kinds of combinations). >> >> I have no experience in comparable setups without the vlan "layer" though. > > I've gotten it to work both with and without vlan just by using > -vlanhwtag. Leaving lro & tso4 enabled works fine. In fact, > everything works fine with vlanhwtag enabled until I throw epair into > the mix. This is using cxgbe/cxl. For the sake of exhaustivity: I have Chelsio cxgb devices in the machine I’m testing this on, but I (am pretty certain) I was only using the Intel igb devices in the aggregate. I am also pretty sure that was without the -vlanhwtag flag that I will now test ASAP.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAD7158B-5481-4323-BBE7-3C9E22308180>