em driver input errors

[Barney Cordoba]

--- On Tue, 9/8/09, Mike Tancsa <mike at sentex.net> wrote:

> From: Mike Tancsa <mike at sentex.net>
> Subject: Re: em driver input errors
> To: "Barney Cordoba" <barney_cordoba at yahoo.com>
> Cc: freebsd-net at freebsd.org
> Date: Tuesday, September 8, 2009, 6:21 PM
> At 05:42 PM 9/8/2009, Barney Cordoba
> wrote:
> > Manish What specific kinds of input errors are you
> getting? How many PPS are you doing, what is the size of the
> ring, and the interrupt modulation rate? Are the NICs PCIe
> or PCIx? Barney
> 
> In my case, our backup server (mix of dump via nfs and some
> dumps through ssh as well as writes via samba mounts) has a
> fairly low pps rate and starts to see input errors at about
> 100Mb.  Adding
> 
> hw.em.rxd=4096
> hw.em.txd=4096
> 
> and
> 
> net.inet.tcp.recvbuf_max=16777216
> net.inet.tcp.recvspace=131072
> net.inet.tcp.sendbuf_max=16777216
> net.inet.tcp.sendspace=131072
> net.inet.udp.recvspace=65536
> kern.ipc.somaxconn=1024
> kern.ipc.maxsockbuf=4194304
> net.inet.ip.redirect=0
> net.inet.ip.intr_queue_maxlen=4096
> net.route.netisr_maxqlen=1024
> kern.ipc.nmbclusters=655360
> 
> didnt seem to make a difference in the amount of errors I
> was seeing
> 
> 
> em1 at pci0:7:5:0: class=0x020000 card=0x348f8086
> chip=0x10768086 rev=0x05 hdr=0x00
>     vendor     = 'Intel
> Corporation'
>     device     = 'Gigabit
> Ethernet Controller (82541EI)'
>     class      = network
>     subclass   = ethernet
>     cap 01[dc] = powerspec 2  supports D0
> D3  current D0
>     cap 07[e4] = PCI-X supports 2048 burst read,
> 1 split transaction
> 
>  Core(TM)2 Quad CPU    Q6600  @ 2.40GHz,
> AMD64, RELENG_7 from Jun 18th
> 
> Plenty of disk IO left and the CPUs doent seem to be taxed
> that much.
> 
> Sep  8 00:02:01 backup3 kernel: em1: Excessive
> collisions = 0
> Sep  8 00:02:01 backup3 kernel: em1: Sequence errors =
> 0
> Sep  8 00:02:01 backup3 kernel: em1: Defer count = 0
> Sep  8 00:02:01 backup3 kernel: em1: Missed Packets =
> 61316
> Sep  8 00:02:01 backup3 kernel: em1: Receive No
> Buffers = 0
> Sep  8 00:02:01 backup3 kernel: em1: Receive Length
> Errors = 0
> Sep  8 00:02:01 backup3 kernel: em1: Receive errors =
> 0
> Sep  8 00:02:01 backup3 kernel: em1: Crc errors = 0
> Sep  8 00:02:01 backup3 kernel: em1: Alignment errors
> = 0
> Sep  8 00:02:01 backup3 kernel: em1: Collision/Carrier
> extension errors = 0
> Sep  8 00:02:01 backup3 kernel: em1: RX overruns =
> 22397
> Sep  8 00:02:01 backup3 kernel: em1: watchdog timeouts
> = 0
> Sep  8 00:02:01 backup3 kernel: em1: RX MSIX IRQ = 0
> TX MSIX IRQ = 0 LINK MSIX IRQ = 0
> Sep  8 00:02:01 backup3 kernel: em1: XON Rcvd = 0
> Sep  8 00:02:01 backup3 kernel: em1: XON Xmtd = 0
> Sep  8 00:02:01 backup3 kernel: em1: XOFF Rcvd = 0
> Sep  8 00:02:01 backup3 kernel: em1: XOFF Xmtd = 0
> Sep  8 00:02:01 backup3 kernel: em1: Good Packets Rcvd
> = 544276980
> Sep  8 00:02:01 backup3 kernel: em1: Good Packets Xmtd
> = 490475071
> Sep  8 00:02:01 backup3 kernel: em1: TSO Contexts Xmtd
> = 0
> Sep  8 00:02:01 backup3 kernel: em1: TSO Contexts
> Failed = 0
> 
> 
> pf is in the kernel as well
> 
> > _______________________________________________
> > freebsd-net at freebsd.org
> mailing list
> > http://lists.freebsd.org/mailman/listinfo/freebsd-net
> > To unsubscribe, send any mail to "freebsd-net-unsubscribe at freebsd.org"
> 
> --------------------------------------------------------------------
> Mike Tancsa,           
>                
>           tel +1 519 651 3400
> Sentex Communications,         
>                
>   mike at sentex.net
> Providing Internet since 1994       
>             www.sentex.net
> Cambridge, Ontario Canada         
>            
>    www.sentex.net/mike
> 
> 

The 8241GI may not be able to handle full gigabit flows if its only
wired at 32-bit 33Mhz, which is only capable of bursting to 1Gb/s. With
a single NIC it likely just fine, but it a bridged or firewall type 
config you may just be seeing bus failures.

Barney