>-----Original Message-----
>From: Gavin Hamill [mailto:[email protected]]
>Sent: Monday, August 09, 2010 4:41 PM
>To: Wyborny, Carolyn
>Cc: [email protected]
>Subject: RE: [E1000-devel] 5% dropped packets under high load
>with e1000e
>
>On Mon, 2010-08-09 at 16:32 -0700, Wyborny, Carolyn wrote:
>
>> > Well, I think flow control isn't the problem. 8-) It can explain
>> dropped packets though. Good question on the ITR setting. That
>> source certainly doesn't look exactly right. I will find
>out for sure
>> and let you know. I will also do some more checking around on
>> possible causes for your symptoms.
>
>Thanks. It's an old version (yay, Debian) so I expect if it's a bug
>it'll have already been fixed long ago.
>
>OK, before I give up for today, I just tried again and this time passed
>ITR params of 0,0,0,0
>
>[39312146.839789] 0000:15:00.0: Interrupt Throttling Rate (ints/sec)
>turned off
>[39312146.995702] eth3: (PCI Express:2.5GB/s:Width x4)
>00:23:7d:fb:df:1a
>
>etc...
>
>and vmstat now shows much higher interrupts as expected:
>
>procs -----------memory---------- ---swap-- -----io---- -system--
>----cpu----
> r b swpd free buff cache si so bi bo in cs us sy
>id wa
> 0 0 60 1291684 164324 347104 0 0 0 0 0 0 1 4 95 0
> 0 0 60 1291676 164324 347104 0 0 0 0 50976 992 0 16 83 0
> 0 0 60 1291312 164324 347104 0 0 0 0 62448 1088 0 20 79 0
> 0 0 60 1291296 164324 347104 0 0 0 0 45185 1015 0 12 87 0
> 0 0 60 1292040 164324 347104 0 0 0 0 34170 1060 1 10 89 0
>
>But... the key thing is a still rising number of dropped connections :(
>
>eth3 Link encap:Ethernet HWaddr 00:24:81:7b:04:ac
> UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
> RX packets:4943354 errors:0 dropped:21757 overruns:0 frame:0
>
>I'm losing hair as it is - I could do without this too ;)
>
>gdh
>
>
>
>
>
Hello Gavin,
After some research, it like your x4 slots might not both be truly x4. A quick
look at your lspci -vvv output looks like one of your pcie slots is only x1.
If you check the dmesg for all four 82571 devices, it does show the width of
the slot its in. I only have the snip of one of them and it shows x4, but I
suspect two of them show x1. Here's what we found in the lspci output:
--snip--
0a:00.0 Ethernet controller: Intel Corporation 82571EB Gigabit Ethernet
Controller (rev 06)
--snip--
LnkSta: Speed 2.5GT/s, Width x1, TrErr- Train- SlotClk+
DLActive- BWMgmt- ABWMgmt-
The 0a:00.1 shows the same thing. If you look down further you'll see that the
other two are really in x4 slots.
Now, I'm wondering if your traffic is distributed equally and just the x1
devices are showing the bottleneck, as would be expected or whether rearranging
devices would help. We have single port devices that have trouble keeping up
in a x1 slot. However, if your traffic is really bunched up on the two
problematic ports, then maybe swapping the hardware around will help.
Let me know if there is more info I can provide.
Thanks,
Carolyn
------------------------------------------------------------------------------
This SF.net email is sponsored by
Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev
_______________________________________________
E1000-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/e1000-devel
To learn more about Intel® Ethernet, visit
http://communities.intel.com/community/wired
