Ethernet Products
Determine ramifications of Intel® Ethernet products and technologies
4811 Discussions

I217-LM huge packetloss with WaitAutoNegComplete=On and FastEthernet/100mbit port

RAren1
Beginner
7,528 Views

We have 30-50% packetloss on a Dell Optiplex 9020 AIO with Intel I217-LM nic, and this renders the machine unusable.

Bios version: A11 (but also tested with A01,A02,A03, will test more, but seems like a driver issue)

OS: Windows 8.1 64bit (but also tested with Windows 7 64bit)

Conditions:

Network interface connected to a FastEthernet port 100mbit with autonegotion on, forcing on 100mbit doesn't fix it.

With Gigabit we don't have any issues.

The issues only appear with the following driver setting:

Devicemanager -> Network adapters -> Intel Ethernet I217-LM -> Properties -> Advanced -> Wait for Link = On

In registry this setting reflects to:

HKLM\SYSTEM\CurrentControlSet\Control\Class\{4d36e972-e325-11ce-bfc1-08002be10318}\0000 WaitAutoNegComplete = 1

If we set this to Off (0) this doesn't happen, but setting to Off brings other issues, because then the driver reports that it has link, and other depending services continue and try to use network.

I made a video to demostrate it:

https://www.youtube.com/watch?v=uNhQEvnUcT0 Intel I217-LM packetloss problems on the Dell Optiplex 9020 AIO - YouTube

This does not happen when the machine is connected to a gigabit switchport.

Tried about 10 different drivers for the I217-LM nic.

Tested on Windows 7 64bit and Windows 8.1 64bit, issue is on both

Any ideas? Does this need fixing from intel's side? Or can this be bios related and does Dell need to release some fix for it?

If it needs fixing from Intel, can this be a new driver, or should it create new firmware.

Can anyone reproduce this? We have about 3000 machines with this NIC so this is a big issue for us.

PS: with the full driver installed there is an extra setting "Test hardware", funny thing is that when we run that whenever the machine is in a "packetloss/broken" state it seems to make it stable again?????

0 Kudos
40 Replies
RAren1
Beginner
896 Views

Hi Rich,

I'm out of the office until monday, but to make it even more difficult to troubleshoot we have multiple Dell 9020 AIO with I217-LM with problems, but also a few without...

Same firmware/bios, same cable, same switchport, same switch, same driver etc.

To rule out software completely we even swapped hd's, but the troublesome one persisted with packetloss. Its only sometimes it seem to initialize correctly (like when that hardware diag is run) without packetloss, but then disabling/enabling nic again results back to the unstable connection.

I will do the Get-IntelNetAdapter | Select * on both Dell 9020 i currently have.

Is there a newer firmware out then 1.3? Is there a way to reflash 1.3 to see if that helps or to check more low level whats the difference between the two?

Thanks! This issue is bugging us for a while, and i wish we could get the funding to just migrate everything to gigabit switches (then the problem goes away) but this is not gonna happen anytime soon at the University...

Reni

0 Kudos
USejd
Beginner
896 Views

Sounds crazy, I know. But please try it:

Connect a usb keyboard or other usb device and check again. Or you can deactivate the USB Root Hub in device manager. I wonder whether you get the same result.

0 Kudos
Richmond_S_Intel
Employee
896 Views

Thanks Reni. I will wait for your update.

Rich

0 Kudos
RAren1
Beginner
896 Views

Rich, sorry for the delay.

I compared both i217-LM's with Get-IntelNetAdapter | Select *, the output is exactly the same, same firmware etc.

 

Grizzlybear for what its worth i also tried fiddling with USB, but this didn't help or change anything.

 

I just got a new system on my desk, the HP model with HP 800 G1 AIO, and it has the same nic, i will do some tests with that system too.

 

Thanks,

 

Reni

 

0 Kudos
RAren1
Beginner
896 Views

Today i tested with the HP 800 G1 AIO with the same interface, and i cannot get the nic into the problematic state.

For what its worth i checked the firmware revision too, but this is also 1.3.

I'm puzzled.....

0 Kudos
st4
New Contributor III
896 Views

Hi Reni,

Thank you for the update. We will check on this.

rgds,

wb

0 Kudos
Richmond_S_Intel
Employee
896 Views

Hi Reni,

I am sorry for the late reply.

So the HP 800 G1 AIO which also uses I217-LM does not exhibit this kind of issue. Can you provide me the device ID?

I might be mistaken but I think you mentioned a while back that you are using a system that has the I218-LM which is not having the same problem, right? Can you also provide the device ID?

Can you try these suggestions provided on the link to see if it helps? http://www.intel.com/support/network/sb/cs-025829.htm Network Connectivity — Tuning Intel® Ethernet Adapter Throughput Performance

Thanks,

 

Rich
0 Kudos
RAren1
Beginner
896 Views

Hi Rich,

>So the HP 800 G1 AIO which also uses I217-LM does not exhibit this kind of issue Can you provide me the device ID??

Correct, DeviceID is: VEN_8086&DEV_153A&SUBSYS_18E6103C&REV4

 

> I might be mistaken but I think you mentioned a while back that you are using a system that has the I218-LM which is not having the same problem, right? Can you also provide the device ID?

True, but the deviceID is the same as the bad ones... So its more a hardware/firmware combination? Although from what i can check the firmware is the same (1.3).

Is there anyway i can try to reflash 1.3 or is there a 1.4 available internally yet which i could try?

PCI\VEN_8086&DEV_153A&SUBSYS_05A71028&REV_04

 

Many thanks,

reni

 

 

 

0 Kudos
st4
New Contributor III
896 Views

Hi Reni,

Thank you for providing the info. We will further look into this.

rgds,

wb

0 Kudos
st4
New Contributor III
896 Views

Hi Reni,

Good day. Further checking this seems to be duplex mismatch issue.Please download the following utility then set I217 to same speed/duplex

as Cisco* to check on the issue:

http://downloadcenter.intel.com/Detail_Desc.aspx?agr=Y http://downloadcenter.intel.com/Detail_Desc.aspx?agr=Y&DwnldID=20585&lang=eng&wapkw=i217

Please contact Dell* support since you have 3000 affected system,it is recommended for them to further assist with the large qty of system involved.

rgds,

wb

0 Kudos
RAren1
Beginner
896 Views

hi wb_intel,

i tried playing with that tool also, but without any success.

It sure sounds like a duplex issue yes, but only with the "Wait for Link" setting active, how is that possible? If you phrase it like that, its more a driver/firmware issue... right?

Many thanks

reni

0 Kudos
st4
New Contributor III
896 Views

Hi Reni,

Thank you for the update. Let me further check.

rgds,

wb

0 Kudos
st4
New Contributor III
896 Views

Hi Reni,

Further checking, please contact Dell* support for further assistance.

thanks,

wb

0 Kudos
RAren1
Beginner
896 Views

Hi wb_intel,

thanks, the unit is actually scheduled today to be picked up by dell.

i will post any outcome to this.

rgds,

Reni

0 Kudos
USejd
Beginner
896 Views

Meanwhile we have found out that the problem appears if on all five usb root hubs the power save option is activated: Allow the computer to turn off this device to save power.

If we deactivate this option on at least one of the usb root hubs, the nic works perfectly. That was the reason why the nic also worked perfectly when we connected at least one usb device to the pc (then power save mode for usb automatically deactivated). It also somehow describes why nic works with 10 MBit/S when no usb device is connected and power save option is activated on all usb root hubs, nic needs less power.

OptiPlex 9020 SFF and HP EliteDesk 800 G1 have also the same manufacturer part: RJ45 jack + 2x USB 2.0 Port.

The EliteDesk 800 G1 that have the same problem have the following part-number from manufacturer UDE: RU1H251NRW2F

http://www.ude-corp.com/frontproduct/product3/lang/en/SmallId/37.html http://www.ude-corp.com/frontproduct/product3/lang/en/SmallId/37.html

0 Kudos
USejd
Beginner
896 Views

Last week the HP came and changed motherboard of our 800 G1, it was exactly the described part (rj-45 jack with 2 usb ports) from UDE that had been replaced on the refurbished motherboard, a spare-part number had been scotched on the described part. No packet loss with the new part / motherboard.

0 Kudos
st4
New Contributor III
896 Views

Hi Grizzlybear,

Thank you for the update and good to know the packet loss did not occur anymore.

rgds,

wb

0 Kudos
RAren1
Beginner
896 Views

On of the Dell systems we had the issue with was send to Dell HQ in the uk and the problem was succesfully replicated there. Awaiting further details on what fix they will provide.

Will report back here ones i have more information.

0 Kudos
st4
New Contributor III
896 Views

Hi Reni,

Thank you for the update

rgds,

wb

0 Kudos
QXiao
Beginner
838 Views

Hi Reni,

can you have some report further?

i have a similar situation.

conditions:

Lenovo Thinkpad E450

win7 + x64bit + I218v + latest dirver, wired with a device(100Mb nic with negotiation).

discripsion:

the connection(show in the bottom right of the Windows desktop) disconnect randomly(maybe few seconds to minute).

windows event log show :

eventId 27 e1dexpress

EventData:

Intel(R) Ethernet Connection (3) I218-V

00040000 00300002 00000000 A004001B

00000000 00000000 00000000 00000000

00000000 00000000 A004001B

do the following tries:

1. disable the shutdown the device to save power(nic property->power management), then not effect

2. change the register key "LinkNegotiationProcess" value to 2(1: AUTO - The Network Adapter will use AutoNegotiation to connect to the link partner at 10 and 100Mbps speed/duplex settings. 2:FORCED - The Network Adapter will force the selected speed and duplex at 10 and 100Mbps settings.) with the "LinkNegProc tool" (which provided by intel) , and no effect

3. do the hardware tests with the Diagnostic option under the tab Link Speed("Register Status", "EEPROM Status", "FIFO Status", "Inerrupt Status" , "Loopback Test(s)"), and no effect.

any ideas ?

best regards.

0 Kudos
Reply