Hey, running a fedora 10 although with a pruned and recompiled kernel for speed.
On an i7 with dual e1000e, at network saturation, I'm seeing this:
eth0 Link encap:Ethernet HWaddr 00:30:48:D7:89:E6
inet addr:192.168.5.35 Bcast:192.168.5.255 Mask:255.255.255.0
inet6 addr: fe80::230:48ff:fed7:89e6/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:123671819 errors:0 dropped:0 overruns:0 frame:0
TX packets:207839589 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:3081243214 (2.8 GiB) TX bytes:3051610475 (2.8 GiB)
Memory:faee0000-faf00000
eth1 Link encap:Ethernet HWaddr 00:30:48:D7:89:E7
inet addr:192.168.7.35 Bcast:192.168.7.255 Mask:255.255.255.0
inet6 addr: fe80::230:48ff:fed7:89e7/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:284927007 errors:0 dropped:48661 overruns:0 frame:0
TX packets:337468721 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:3792748252 (3.5 GiB) TX bytes:25071278 (23.9 MiB)
Memory:fafe0000-fb000000
Note the drops on eth1. (It is running more load than eth0, especially small
packet wise).
It is setting up and tearing down around 110,000 tcp connections a second.
Eth1 doesn't have a shared interrupt, note that I was experimenting with
irqbalanced on and off so the below
may look wierd.
[r...@wrjo8e1i bin]# cat /proc/interrupts
CPU0 CPU1 CPU2 CPU3
0: 124 0 0 9885021 IO-APIC-edge timer
1: 0 0 0 2 IO-APIC-edge i8042
6: 0 3 0 0 IO-APIC-edge floppy
8: 0 0 1 0 IO-APIC-edge rtc0
9: 0 0 0 0 IO-APIC-fasteoi acpi
12: 4 0 0 0 IO-APIC-edge i8042
16: 0 874993 8128708 1375934 IO-APIC-fasteoi
uhci_hcd:usb3, eth0
17: 0 27954150 8523866 0 IO-APIC-fasteoi eth1
18: 0 0 0 0 IO-APIC-fasteoi
ehci_hcd:usb1, uhci_hcd:usb6
19: 11 433 3706 5145 IO-APIC-fasteoi ata_piix,
ata_piix, uhci_hcd:usb5
23: 0 0 0 0 IO-APIC-fasteoi
ehci_hcd:usb2, uhci_hcd:usb4
NMI: 0 0 0 0 Non-maskable interrupts
LOC: 2704515 3443362 3598215 789581 Local timer interrupts
RES: 3803504 2660844 3622263 3830197 Rescheduling interrupts
CAL: 87 98 89 29 function call interrupts
TLB: 13008 37360 24688 13609 TLB shootdowns
TRM: 0 0 0 0 Thermal event interrupts
SPU: 0 0 0 0 Spurious interrupts
ERR: 0
I tried irqbalanced on and off, I tried ethtool -G eth1 rx 4096, and I tried
sysctl -w net.core.netdev_max_backlog=4096
and still get drops.
Linux wrjo8e1i 2.6.27.21-170.ns5.fc10.i686 #1 SMP Thu Apr 16 15:58:56 EDT 2009
i686 i686 i386 GNU/Linux
[r...@wrjo8e1i bin]# modinfo e1000e
filename:
/lib/modules/2.6.27.21-170.ns5.fc10.i686/kernel/drivers/net/e1000e/e1000e.ko
version: 0.3.3.3-k6
license: GPL
description: Intel(R) PRO/1000 Network Driver
author: Intel Corporation, <[email protected]>
0000:06:00.0: eth0: (PCI Express:2.5GB/s:Width x1) 00:30:48:d7:89:e6
0000:06:00.0: eth0: Intel(R) PRO/1000 Network Connection
0000:06:00.0: eth0: MAC: 3, PHY: 8, PBA No: 0101ff-0ff
0000:07:00.0: eth1: (PCI Express:2.5GB/s:Width x1) 00:30:48:d7:89:e7
0000:07:00.0: eth1: Intel(R) PRO/1000 Network Connection
0000:07:00.0: eth1: MAC: 3, PHY: 8, PBA No: 0101ff-0ff
e1000e: Intel(R) PRO/1000 Network Driver - 0.3.3.3-k6
e1000e: Copyright (c) 1999-2008 Intel Corporation.
e1000e 0000:06:00.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
e1000e 0000:06:00.0: setting latency timer to 64
e1000e 0000:07:00.0: PCI INT A -> GSI 17 (level, low) -> IRQ 17
e1000e 0000:07:00.0: setting latency timer to 64
------------------------------------------------------------------------------
Stay on top of everything new and different, both inside and
around Java (TM) technology - register by April 22, and save
$200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco.
300 plus technical and hands-on sessions. Register today.
Use priority code J9JMT32. http://p.sf.net/sfu/p
_______________________________________________
E1000-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/e1000-devel
