I found an issue and I've narrowed it down to the Intel driver.

Just to show the work I've put into this:

- I used iperf to test the bandwidth on all server types (Physical Linux install and inside a VM)

- I have an existing VMware support ticket and we've worked it down to the driver.

- I tested the issue with three different x540-T2 adapters and a x520-DA2 adapter.

Problem: When a server with the intel x540-T2 adapter is connected at 10Gbps and tries to transmit to another server that is connected at 1Gbps, All flows out of the 10Gbps adapter drops to a max of 1Gbps until the 1Gbps flows stop.

Let me clarify a bit:

Server A has an Intel x540-T2 adapter and is connected at 10Gbps.

Server B has an Intel x540-T2 adapter and is connected at 10Gbps.

Server C has a HP Server adapter and is connected at 1Gbps.

I start a Transmit flow from Server A to Server B and I see the iperf session reach full line rate (10Gbps)

I start a concurrent iperf flow From server A to Server C.

My expectation is that Server A will continue to transmit at 10Gbps (total), Server B will drop down 1Gbps to (8-9Gbps) to allow bandwidth for the 1Gbps flow to Server C.

BUT, this does not happen.

Server A will flop down to 1.5Gbps transmit and Server B and Server C will now receive 500-700Mbps until the flow to Server C stops then the flow from Server A to Server B goes back to 10Gbps.

Driver data:

If I use driver version 3.7.13.7  this problem does not exist!  Everything works as expected and Server A continues to transmit at 10Gbps while Server B drops to 8Gbps+ and Server C receives at a full 1Gbps.

If I Update to any version after that, it has the transmit flop.

Driver Version 3.7.13.7 is the default driver used in the ESXi 5.5 build 1331820 install ISO.

Newer version tested: (these are listed as compatible with ESXi 5.5 on the VMware compatibility guide)

3.15.1.8

3.18.7

3.19.1

Some extra info from ESXi cli:

###### This is the working Driver:########

~ # vmkchdev -l | grep vmnic
0000:01:00.0 8086:1528 8086:001a vmkernel vmnic0
0000:01:00.1 8086:1528 8086:001a vmkernel vmnic1

~ # ethtool -i vmnic0
driver: ixgbe
version: 3.7.13.7.14iov-NAPI
firmware-version: 0x80000389
bus-info: 0000:01:00.0

~ # esxcfg-nics -l
Name    PCI           Driver      Link Speed     Duplex MAC Address       MTU    Description
vmnic0  0000:01:00.00 ixgbe       Up   10000Mbps Full   a0:36:9f:01:b6:28 9000   Intel Corporation Ethernet Controller X540-AT2

###### This is one of the broken Driver versions:########

~ # vmkchdev -l | grep vmnic
0000:01:00.0 8086:1528 8086:0001 vmkernel vmnic0
0000:01:00.1 8086:1528 8086:0001 vmkernel vmnic1

~ # ethtool -i vmnic0
driver: ixgbe
version: 3.15.1.8iov
firmware-version: 0x80000389
bus-info: 0000:01:00.0

~ # esxcfg-nics -l
Name    PCI           Driver      Link Speed     Duplex MAC Address       MTU    Description
vmnic0  0000:01:00.00 ixgbe       Up   10000Mbps Full   a0:36:9f:01:b6:28 9000   Intel Corporation Ethernet Controller X540-AT2

Things to note here are the VID and DID are still being detected correctly as 8086 and 1528.

Any thoughts or questions?

I have a ton of screen shots if needed to prove the case.

Hi, I have recently bought a i350-T4 NIC, and flash the iscsi boot agent into it (18.8 version)

After reboot , the rom said   "error : no iscsi boot port available" and don't have the ctrl +d to do configuration.

search on the internet but cannot find any information about this problem.

i tried to use the intel CT adapter before and don't have this problem.

just press ctrl+ D to configure and connect to the iscsi target.

Could you please help me resolve this issue?

Thanks

NIC : i350-T4

MB : intel s1200btsr

CPU : Xeon E1200-V2

I'm not sure if this is good category. If not- im sorry.

Hi,

Couple days ago i just bought a new computer (Asrock z87 extreme3, intel i5-4670k...). So i buit it by myself, everything went alright, settings, then i installed windows 7 x64. So then was time for drivers which i installed from CD, i was really happy, until i plug in rj-45 into. Connection is limited to 2 Mbps download and 0.03 Mbps upload, which i have connection 30 mbps / 3Mbps. Router is Linksys wrt54GL which have nothing to do, becouse when i unplug cable and plug into my laptop (hp g62-b45ew Win 7 x86) i have connection 30/3 Mbps.

I was using drivers from attached disc, so i download the newest from asrock site. I remove old drivers, install new, and nothing have change. I also download the newest drivers from intel site ( its internet adapter intel ethernet connection I217-V)- nothing have change.

What should i do ? Is it becouse im using win7 instead of win8 ? I really don't want to change os and make format.

Here are prove:

1. Cable in laptop:

http://static.pokazywarka.pl/i/2654268/796494/2.jpg

2. The same cable, on the same settings on new computer:

http://static.pokazywarka.pl/i/2654274/326518/1.jpg

Operating system: Windows 7 SP1 x64.

Hardware: Hp Elite 8300 CMT Desktop with Integrated Intel® 82579LM GbE Network Connection (supporting vPro remote management technologies).

The Question?

Teaming can be applied between  Intel Gigabit CT PCI-E and the built in NIC mentioned above?

or shall be 2 NIC (Intel Gigabit CT PCI-E)?.

I ordered the 10 PCs and i am confused in ordering the NIC shall it be 10 or 20?

my application requires  dual lan support teaming to setup one ip and use the load balancing,link aggregation...etc.

Hi,

my issue seems to be related to Intel 82579LM Gigabit Ethernet - UDP Packet Loss, but in my case, small TCP packets get lost every couple of seconds.

In my case, a DELL Latitude E6520 notebook communicates with a WAGO I/O-System device via Modbus TCP at a "high" rate of small packets (TCP_NODELAY), i. e. around 1000 packets per second (9.58 % of size 66 byte on wire, 90.42 % of size 117 or 119 byte on wire according to Wireshark).

Every couple of seconds, it happens that a reply of the WAGO device doesn't appear on the notebook, causing a delay of magnitude 3 compared to the normal packet exchange, which severely impacts the hardware controlled by the WAGO device.

Notebook and WAGO device are connected by a D-Link DGS-1210-16 switch. Activating port mirroring on the notebook port and running Wireshark on an different PC connected to the mirror port reveals, that the reply of the WAGO device is actually sent to the notebook, but the notebook doesn't see it and therefore the TCP stack on the notebook requests a retransmission after a delay of about 1 second.

I currently work around this issue by mirroring the sent packets of the WAGO device to the notebook port on the switch.

The notebook is running with current BIOS and is connected to the mains. Recent Intel network drivers are installed, using default settings. Diagnostics doesn't reveal anything bad.

What can I further do, to get that issue fixed?

Thanks in advance. Bye.

Hello everyone.

After cloning MAC adress on my Ethernet adapter Intel 82577LC, the error PXE-E05 occured (in OS windows 8.1 the ethernet adapter was with yellow exclamation mark and led on the adapter had blinked).

I found in the internet the solution, using IBAUTIL in MS-DOS mode to set default configuration .

After setting to default by using "IBAUTIL -NIC=1 -devcfg" my adapter is down.

No one LED blinks and in the device manager is not found (only wifi is found).

I tried once again to use IBAUTIL, but there write error "No supported network adapters found".

In BIOS I tried to set diffirent configuration, defaults settings, but still the network adapter is not found.

I tried also to remove battery for 10 minut, but nothing happened.

Could you mind help me with this problem.

After installing this driver i started creating the vlans: 2 tagged vlans and 1 untagged vlan.

the untagged vlan is out primary vlan used to connect to our servers and internet.

This primary vlan didn't seem to work properly.. I could connect to the internet but not to all local servers.

When looking at the traffic flow on the server i was connecting to i saw traffic being send from the interface and toward the interface but it never reaches the browser. when reverting to version 18.8 it all started to work again.

further it is a pain to find a place where to file a bug report on this site..

I have a new intel pro/1000 MT server adapter, adter so many problems to get it work (both on ubuntu 13.10 and windows), there seems to be a major issue in it's operation. when using cat6 patch cords (tried different 4 new patch cords all manufactured by Belden, and all tested to be OK by other adaprtos ) there is a huge packet loss, as much as the interface speed is in range of kbps when connecting to another GB interface, just by replacing the cable with any old cat5e, it would work fine! what could be the cause?!

let me repeat that I have checked those CAT6 cables, connecting other interfaces, and they work perfectly, the issue is only happening for this 82545gm intel pro/1000mt interface. as much as I know CAT6 is fully backward compatible! so I cant understand whats wrong in here! int

Update:

after checking the interface statistics I see many errors:

ethtool -S eth1

     rx_errors: 1670
     tx_errors: 0
     tx_dropped: 0
     multicast: 0
     collisions: 0
     rx_length_errors: 0
     rx_over_errors: 0
     rx_crc_errors: 9354

but I don't know the reason!

Message was edited by: hasan m

I am testing RHEL 6.4 VMs (ixgbevf 2.6.0-k driver) on RHEL 6.5 Hypervisor (KVM with the ixgbe3.15.1-k driver) using SR-IOV .  I have a high-rate multicast application and it appears that every VM is receiving the packets for every multicast packet going into the PF, whether the VM associated with the VF registered for that multicast address or not.  Does anyone know if we have something misconfigured?    Is there anyway to configure the PF to only forward the multicast packets to VFs that registered for particular multicast addresses?

Thanks!

Hi everyone,

Intel 82576 datasheet says it has header split feature,

but I can't find any enable/disable header split option in device management,

I download driver is the newest, Can anyone help me for how to enable header split for Intel 82576?

Thank you for help.

Hello,

I ran into the problem that wake on lan is not working with freebsd and intel nic.

Shutdown the system with poweroff, WOL is not working.

if I comment out

/sbin/ifconfig -l | /usr/bin/xargs -n 1 -J % /sbin/ifconfig % down

in /etc/rc.shutdown, wake on lan is working.

the catch - after the reboot, the interface doesn't react (no IP - not even static - can be assigned). I've to poweroff the machine again (with ifconfig down) for the interface to work again.

Port LEDs stay on - with and without ifconfig down.

Thanks,

Andreas

Here are the details:

FreeBSD 9.2-RELEASE-p3

em0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500

        options=4219b<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,VLAN_HWCSUM,TSO4,WOL_MAGIC,VLAN_HWTSO>

        ether 00:15:17:31:XX:XX

        inet 172.17.0.6 netmask 0xffffff00 broadcast 172.17.0.255

        nd6 options=9<PERFORMNUD,IFDISABLED>

        media: Ethernet autoselect (1000baseT <full-duplex>)

        status: active

em0: <Intel(R) PRO/1000 Network Connection 7.3.8> port 0xbc00-0xbc1f mem 0xff6e0000 0xff6fffff irq 16 at device 0.0 on pci1

em0: Using an MSI interrupt

em0: Ethernet address: 00:15:17:31:XX:XX

Hi,

I have installed PROWinx64.exe. in Windows 7 64 bits and trying to set speed of all the installed Intel NICs in the PC using PROSETCL. I use the Adapter_EnumerateSettings 3 command to get the settings of the NIC. I see *SpeedDuplex             - Auto Negotiation.

When i try to get the setting using: PROSetCL.exe Adapter_GetSetting 3 *SpeedDuplex

I get the following Error: Setting name not found: *SpeedDuplex

Does anyone know why it Settcan not find the setting name, although, is right.

Thanks.

I have an otherwise functioning install of Windows 7 64bit edition. Sometimes, the process NT Kernel & system jumps to 20-30% from its usual 0. I can sometimes go a week without it happening. Sometimes, it's an hour. It goes away if I reboot. It also goes away permanently if I disable the 82579LM network card. So, that's the problem.

I've tried the Dell drivers, the last 2 or 3 versions of Intel drivers, and it still occurs. I've also tried virus and malware scans. Anybody have an idea how to fix it? It's not a huge deal, but I'd still like to fix it.

Can anyone tell me where I can obtain a BSDL file for the an 82572EI Ethernet Controller?  I can't find it as a download. The device claims to support boundary scan testing so presumably the BSDL must exist?! Thanks.

Hello all. I am unable to disable the "Reduce link speed to save power" and "Reduce link speed during standby" options under the Power Management tab, in the Intel 82566DC Gigabit Network Connection Properties. This occurs in Windows XP Professional x64 Edition (SP2) with Intel PROSet Version: 18.3.62.0. I cannot disable these settings either in OS Normal or Safe Mode. Is this by design?

On the same computer, I also have Windows XP Home Edition (SP3) installed on a different partition. Under this setup, the Intel PROSet version is older, but I am able to disable and enable the same applicable power settings.

Thank you all for your helpful and timely responses.

Specs:

Dell Dimension 9200 DXP061,

Intel DualCore CPU @ 2.4 GHz,

4 GB RAM.

When I send a pause frame on my 10G SFP+ SuperMicro Board (INTEL chipset), I cannot see the pause frame detected on wireshark. The Linux driver is ixgbe.

When I send a pause frame on my 10G CX4 board, I CAN See the pause frame detected on wireshark.

Can you tell me why I cannot see the pause frame with the 10G SFP+ SuperMicro Board on wireshark.

Thank you for help. Serge Lin

Hello,

I have an Intel X540-T1 NIC installed in one of our servers that is exhibiting the problem outlined in the FreeBSD problem report kern/183390.  The link to the problem report is as follows....

kern/183390: [ixgbe] 10gigabit networking problems

We are using the latest available version of the driver (2.5.15).  I'm wondering when a newer version of the FreeBSD driver for this NIC will be released that resolves the issue described in the previously mentioned problem report.  Basically, the NIC stops responding after moderate to heavy traffic load and the only way to solve the problem is to bring the interface down and then back up.  I look forward to any suggestions about this issue.  And hope Intel will take a good look at the problem and release a new version of the driver that solves it.  Thank you.

Regards,

Eric

Hi,

We have a number of Intel 82599EB 10-Gigabit dual port NICs running in a number of servers.  On a couple of these servers, we are seeing very strange behaviour where when connecting via a Brocade 1G/10G switch from a 1G server we are getting line speed RX (113MB/s) but only a fraction of this TX, for example:

# ./netperf -H XXX.XXX.XXX.XXX -f M
MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to XXX.XXX.XXX.XXX () port 0 AF_INET
Recv   Send    Send
Socket Socket  Message  Elapsed
Size   Size    Size     Time     Throughput
bytes  bytes   bytes    secs.    MBytes/sec

2097152  65536  65536    10.14     113.16

# ./netperf -H XXX.XXX.XXX.XXX -f M -t TCP_MAERTS
MIGRATED TCP MAERTS TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to XXX.XXX.XXX.XXX () port 0 AF_INET
Recv   Send    Send
Socket Socket  Message  Elapsed
Size   Size    Size     Time     Throughput
bytes  bytes   bytes    secs.    MBytes/sec

87380 2097152 2097152    10.00      20.38

We have another server with the same card and that behaves as expected (line speed both TX and RX).  As far as I can tell all network configs are the same on both boxes.  We've tried different switch, cable, OS and server hardware.

Am I missing something really obvious here?!

Thanks,

Chris

Hello,

The PRO/1000 PF Dual Port Server Adapter ignores the content of the 2-octet length field in the Ethernet frame. Its successor, the Ethernet Server Adapter I350-F2, however, does care about it. If the actual packet length exceeds it, the card seems to simply drop the packet (the packet doesn't make it to the driver).

Is it possible to make the i350 ignore this field just as the old PRO/1000 did? I'd like it not to drop the suspicious packets.

Thank you.

Karoly

Hi,

We have a Supermicro server with 10G Card (82599EB 10-Gigabit SFI/SFP+) and ixgbe version 3.18.7.  We are noticing something strange happening with the networking service. The service is going down randomly without any cause. The issue can recreated by running ethtool -t <eth>

Found the following on the logs:

Mar 20 23:41:59 <host> kernel: [ 2792.727776] ADDRCONF(NETDEV_UP): eth1: link is not ready

Mar 20 23:42:03 <host> kernel: [ 2796.802635] ixgbe 0000:05:00.0: eth1: detected SFP+: 3

Mar 20 23:42:04 <host> kernel: [ 2796.926539] ixgbe 0000:05:00.0: eth1: NIC Link is Up 10 Gbps, Flow Control: RX/TX

Mar 20 23:42:04 <host> kernel: [ 2796.927353] ADDRCONF(NETDEV_CHANGE): eth1: link becomes ready

Mar 20 23:42:14 <host> kernel: [ 2807.512585] eth1: no IPv6 routers present

When we debugged further, we found that the default route is getting disappeared somehow and shows up only after a network restart. From this forum , we suspected that "LRO " is causing the issue. We disabled it by:

ethtool -K <eth> lro off

which didn't fix. I also tried compiling ixgbe with 'CFLAGS_EXTRA="-DIXGBE_NO_LRO"' which also didnt help.

Can someone help with this case?

Thanks,

--

Vaisakh