Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 20 Aug 2009 05:34:38 -0700 (PDT)
From:      alexpalias-bsdnet@yahoo.com
To:        Dmitriy Zamuraev <gigabyte.tmn@gmail.com>
Cc:        freebsd-net@freebsd.org
Subject:   Re: em driver input errors
Message-ID:  <487959.75728.qm@web56402.mail.re3.yahoo.com>
In-Reply-To: <010901ca218d$878276a0$1e010a0a@in72.ru>

next in thread | previous in thread | raw e-mail | index | archive | help
=0A=0A--- On Thu, 8/20/09, Dmitriy Zamuraev <gigabyte.tmn@gmail.com> wrote:=
=0A=0A> From: Dmitriy Zamuraev <gigabyte.tmn@gmail.com>=0A> Subject: Re: em=
 driver input errors=0A> To: alexpalias-bsdnet@yahoo.com=0A> Cc: freebsd-ne=
t@freebsd.org=0A> Date: Thursday, August 20, 2009, 2:58 PM=0A> Hello Alex,=
=0A> =0A> > SCHED_ULE, HZ=3D1000:=0A> I use this too=0A> =0A=0A> >  And "ne=
tstat -ni"=0A> shows 0 errors on all cards?=0A> Not exactly zero, but for u=
ptime 155 days it seems to be=0A> ok.=0A> bras1 [/usr/home/dm]# netstat -i|=
grep em=0A> em0=A0 =A0 1500 <Link#1>=A0 =A0 =A000:15:17:71:f8:52 2457503820=
 20175 2096211799 0=A0=A00=0A> em1=A0 =A0 1500 <Link#2>=A0 =A0 =A000:15:17:=
71:f8:52 1084492221 11188 909418060=A0=A0=A0=A00 0=0A> em2=A0 =A0 1500 <Lin=
k#3>=A0 =A0 =A000:15:17:71:f8:52 4212941427 29566 3500442287 0=A0=A0=A0=A00=
=0A> em3=A0 =A0 1500 <Link#4>=A0 =A0 =A000:15:17:71:f8:52 2143321197=A0 =A0=
=A0=A00 1878792786 0=A0 =A0=A0=A00=0A=0ADefinitely way better than my numbe=
rs.=0A=0AName    Mtu Network       Address         Ipkts Ierrs    Opkts Oer=
rs  Coll=0Aem0    1500 <Link#1>      blah1    11166666892 2776856 131945601=
32     0     0=0Aem1    1500 <Link#2>      blah2    7420822703 462462 61850=
54971     0     0=0Aem2    1500 <Link#3>      blah3    6070962640 167193 53=
41606634     0     0=0Aem3    1500 <Link#4>      blah4    15183452     0 17=
424318     0     0=0A=0A# uptime=0A 3:29PM  up 8 days, 11:19, 2 users, load=
 averages: 0.69, 0.78, 0.83=0A=0A=0A> This counters was made by UDP flood, =
when dummynet can't=0A> process all packets and=0A> swi:net loads one core =
up to 100%. Yes, the dummynet on=0A> this machine, its bad idea but it's=0A=
> working stable. (After this incident the switches now=0A> control the flo=
od attack)=0A> NOTE: the MAC is equal because i use if_lagg(4) for this=0A>=
 interfaces for load all cores in CPU=0A=0AInteresting.  But I'm seeing thi=
s during normal traffic.  I can't remember if it was this machine, or the o=
lder machine (also with em driver) when I did get a flood of over 90.000 pa=
ckets/second, and the error rate jumped to around 50.000/second...=0A=0A> >=
> I think it depends by motherbord, what full=0A> hardware specification ar=
e you using? with chips names=0A> > The machine is a Dell PowerEdge 2850.=
=A0 According=0A> to its specs, the chipset is Intel E7520.=0A> > Two 64-bi=
t Xeon processors at 3.20GHz, 4 GB RAM.=0A> For you bandwidth this server m=
ust work fine.=0A> Check the UDP/ICMP or other flood on em0 when errors=0A>=
 appear.=0A> What kind of device at the end of em0 copper cable?=0A> If thi=
s a manageable switch, and supports tools - try to=0A> investigate what hap=
pens when errors appear.=0A=0AIt's a ZyXEL managed switch, unfortunately SN=
MP support doesn't seem to work very well on it.  But even before I added t=
his switch (when there was a Cisco Catalyst 3560 at the end of the cable) I=
 was seeing the error bursts.=0A=0ABut I will try and see if I detect anyth=
ing on the switch end.=0A=0AThanks=0AAlex



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?487959.75728.qm>