Date: Fri, 21 Mar 2014 14:22:18 -0300 From: Christopher Forgeron <csforgeron@gmail.com> To: Markus Gebert <markus.gebert@hostpoint.ch> Cc: FreeBSD Net <freebsd-net@freebsd.org>, Rick Macklem <rmacklem@uoguelph.ca>, Jack Vogel <jfvogel@gmail.com> Subject: Re: 9.2 ixgbe tx queue hang Message-ID: <CAB2_NwDJ%2BCcEkSZjnPYU_Rg-gSCkkfHtuWYiyEixAvhORLMo7g@mail.gmail.com> In-Reply-To: <7B586D3A-BD6B-40B1-980E-7F9FD4A49F6A@hostpoint.ch> References: <CAB2_NwB=21H5pcx=Wzz5gV38eRN%2BtfwhY28m2FZhdEi6X3JE7g@mail.gmail.com> <1543350122.637684.1395368002237.JavaMail.root@uoguelph.ca> <CAB2_NwCGsAHdMFPoST05azb9K_O-K_khk3Bi1sF2om3puCcyCw@mail.gmail.com> <CAB2_NwC3on1xP3UAutkQa-3zu_JhK0%2B-ZjVb6_3NVemw2Or-KQ@mail.gmail.com> <CAB2_NwBSc3KWPYD-xbWYpRFTxpsKnXEr4V1ySP5g83aZM59MvQ@mail.gmail.com> <D1B4320A-DFFD-4647-8A43-238A088D7EF1@hostpoint.ch> <CAB2_NwAq-5N53hVPxOqc__kBi_QCQ0iy4fiwQJW0FTRqueN0YQ@mail.gmail.com> <34491192-F8EC-45C1-A7C8-61C3EBE5CFBD@hostpoint.ch> <CAB2_NwCYtFD1hkYwO-aC%2BuyUGjU_UopExsc0Kq8afuefiSdGtA@mail.gmail.com> <7B586D3A-BD6B-40B1-980E-7F9FD4A49F6A@hostpoint.ch>
next in thread | previous in thread | raw e-mail | index | archive | help
Fair enough. Have you tried disabling tso on the ix's ? That does fix the problem for me, however there is a performance penalty to be paid. I'm now regressing through the ixgbe drivers - I see there's been changes to how the queues are drained between 9.1 - 10.0, will see if the older ixgbe 2.4.8 works under 10.0 On Fri, Mar 21, 2014 at 1:40 PM, Markus Gebert <markus.gebert@hostpoint.ch>wrote: > > On 21.03.2014, at 15:49, Christopher Forgeron <csforgeron@gmail.com> > wrote: > > > However, if you can make a spare tester of the same hardware, that's > > perfect - And you can generate all the load you need with benchmark > > software like iometer, large NFS copies, or perhaps a small replica of > your > > network. Synthetic load is easier to control, and thus easier to > reproduce > > the issue and speed testing. Heck, you may be able to do it all by > looping > > through your two ix adapters and never using an external client. > > > > It's a bit of a pain to setup, but it's worth the effort imo. > > The main problem is, that all the affected systems are blades which are > only connected 1gig. I think that's the main reason we have trouble > reproducing the problem and I cannot change that, because we simply lack > the parts to produce any kind of 10gig connection between blades. So I will > postpone this idea, especially since our problems seem very similar again. > 9.2 or 10.0 does not seem to matter, at least for now. > > > Markus > >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAB2_NwDJ%2BCcEkSZjnPYU_Rg-gSCkkfHtuWYiyEixAvhORLMo7g>