On Fri, Oct 7, 2011 at 5:23 PM, Brandeburg, Jesse
<[email protected]> wrote:
>
>
> On Thu, 6 Oct 2011, Jesse Gross wrote:
>
>> I'm seeing some strange packet reordering problems with Intel 82599
>> based NICs using the ixgbe driver.  It seems difficult to believe that
>> I'm the only one running into this but it has shown up to some extent
>> with every card of this type that I have tried, on multiple systems,
>> using multiple kernels.  By contrast, I also tried Broadcom-based 10G
>> NICs and did not see the problem in any of these environments.
>>
>> I tried to collect information on the simplest, most common setup - I
>> have two machines running RHEL 6.1 connected back-to-back using SFP+
>> direct attach cables.  I changed nothing after installation except for
>> installing netperf.  I then ran a basic test without any other
>> traffic:
>
> with that statement I think I might see your problem.  With the default
> setup irqbalance is enabled and might be migrating your interrupts
> frequently, which causes out of order reception.
>
> could you try killall irqbalance and run the test again?  If that works I
> recommend (as our readme does/should) that you run the set_irq_affinity.sh
> script from the sourceforge ixgbe tarball to lay your interrupts out in
> an efficient manner.  plus see below.

irqbalance was running but after killing it I didn't see any
appreciable difference.  I also tried running set_irq_affinity.sh for
the fun of it with the same results.

>> TcpExt:
>>     2 delayed acks sent
>>     23 packets directly queued to recvmsg prequeue.
>>     114 packets header predicted
>>     745471 acknowledgments not containing data received
>>     143904 predicted acknowledgments
>>     3 times recovered from packet loss due to SACK data
>>     Detected reordering 319 times using SACK
>>     Detected reordering 1124 times using time stamp
>>     3 congestion windows fully recovered
>>     1124 congestion windows partially recovered using Hoe heuristic
>>     0 TCP data loss events
>>     2165 fast retransmits
>
> the retransmits here might also be because you're overwhelming the
> receiver and causing dropped packets either in the driver or the socket
> layer.

It seems somewhat unlikely to me that the machine is too slow since I
don't see this problem with 10G Broadcom NICs.  Looking at mpstat, the
CPU that traffic is being directed to is about 60% idle (and
everything else is completely idle).  These particular machines have
an Intel E5520 in them, although I have also seen results like this
with slightly faster machines as well.

------------------------------------------------------------------------------
All of the data generated in your IT infrastructure is seriously valuable.
Why? It contains a definitive record of application performance, security
threats, fraudulent activity, and more. Splunk takes this data and makes
sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-d2dcopy2
_______________________________________________
E1000-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/e1000-devel
To learn more about Intel&#174; Ethernet, visit 
http://communities.intel.com/community/wired

Reply via email to