Date: Thu, 20 Aug 2009 03:07:54 -0700 (PDT) From: alexpalias-bsdnet@yahoo.com To: =?utf-8?B?0JTQvNC40YLRgNC40Lkg0JfQsNC80YPRgNCw0LXQsg==?= <gigabyte.tmn@gmail.com> Cc: freebsd-net@freebsd.org Subject: Re: em driver input errors Message-ID: <550146.64358.qm@web56404.mail.re3.yahoo.com> In-Reply-To: <000e01ca20e9$e19caa10$1e010a0a@in72.ru>
next in thread | previous in thread | raw e-mail | index | archive | help
Hello.=0A=0A--- On Wed, 8/19/09, =D0=94=D0=BC=D0=B8=D1=82=D1=80=D0=B8=D0=B9= =D0=97=D0=B0=D0=BC=D1=83=D1=80=D0=B0=D0=B5=D0=B2 <gigabyte.tmn@gmail.com> = wrote:=0A=0A> From: =D0=94=D0=BC=D0=B8=D1=82=D1=80=D0=B8=D0=B9 =D0=97=D0=B0= =D0=BC=D1=83=D1=80=D0=B0=D0=B5=D0=B2 <gigabyte.tmn@gmail.com>=0A> Subject: = Re: em driver input errors=0A> To: alexpalias-bsdnet@yahoo.com=0A> Cc: free= bsd-net@freebsd.org=0A> Date: Wednesday, August 19, 2009, 7:27 PM=0A> Hello= Alex.=0A> =0A> What sheduler are you using? ULE or 4BSD=0A> Have you NIC I= RQ sharing with other hardware?=0A> What HZ value? 1000?=0A=0ASCHED_ULE, HZ= =3D1000:=0A=0Ahost# sysctl kern.sched.name kern.hz=0Akern.sched.name: ULE= =0Akern.hz: 1000=0Ahost#=0A=0A> > Thanks for the suggestion.=0A> > From a "= clean" box:=0A> > dev.em.0.rx_int_delay: 0=0A> > dev.em.0.tx_int_delay: 66= =0A> > dev.em.0.rx_abs_int_delay: 66=0A> > dev.em.0.tx_abs_int_delay: 66=0A= > > I reset all the values (errors still appearing), then=0A> tried your su= ggestion (rx_int_delay=3D600,=0A> rx_abs_int_delay=3D1000).=C2=A0 This has = reduced the number of=0A> >interrupts for em0 (from about 7200/sec to aroun= d=0A> 6500/sec).=C2=A0 After some time, I started getting errors=0A> again.= =0A> mmm, try the maximum value 67108, what hapens...=0A=0AI will try this = today when there's enough traffic to see errors.=0A=0A> > But that has made= me try this also:=0A> > dev.em.0.tx_int_delay=3D600=0A> > dev.em.0.tx_abs_= int_delay=3D1000=0A> I think it's a bad idea, but don't know because:=0A> >= Meaning using your suggested values for tx too.=C2=A0=0A> Now em0 is seein= g about 1800 interrupts/second, which is way=0A> better, but after some tim= e I saw errors >again...=0A> =0A> > From the output of "netstat -nI em0 -w = 5":=0A> maybe mistake, did you meen "netstat -w5 em0" ?=0A=0ANope, exactly = as in my mail, "netstat -nI em0 -w 5". It does take 5 seconds to produce m= eaningful output.=0A=0A> I have PPPoE concenrator based on S3000AHV motherb= oard with=0A> Core2Quad 6600 and four (to load all cores in CPU) Intel=0A> = PCI-E x1 and PCI-E x4 NIC's=0A> My load:=0A=0APretty impressive figures. A= nd "netstat -ni" shows 0 errors on all cards?=0A=0A> And i have't any probl= ems. I think i select the good=0A> hardware.=0A> =0A> > Interrupts total (a= s reported by systat):=C2=A0 around=0A> 13500/second.=C2=A0 I would estimat= e the old IRQ load at=0A> around 30000-35000/second, which doesn't seem too= >much=0A> to me, for a dual xeon machine.=0A=0A> I think it depends by mot= herbord, what full hardware=0A> specification are you using? with chips nam= es=0A=0AThe machine is a Dell PowerEdge 2850. According to its specs, the = chipset is Intel E7520. Two 64-bit Xeon processors at 3.20GHz, 4 GB RAM.= =0A =0A> > Speaking of which, I did compile the kernel with=0A> "options DE= VICE_POLLING", but enabling polling only made the=0A> errors appear more of= ten, and in greater >numbers.=0A> I don't use polling on FBSD 7.x, it's usa= ble on FBSD older=0A> versions=0A=0AI tried as many possibilities as I coul= d.=0A =0A> > - 1 x dual-port gigabit interface, PCI-X=0A> Maybe I have this= card. And it works unstable, i don't=0A> remember what happens, but i seen= by tcpdump "truncated IP,=0A> missing XX bytes"=0A=0ACurrently most errors= are on the motherboard-embedded em0 interface. Second is embedded em1. L= ast are em2 and em3 which are on the dual-port card (em2 under 170k errors = as opposed to 2.6M for em0, and em3 0 errors)=0A> Good luck. =0A=0AThanks!
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?550146.64358.qm>