MAC addresses differ because they tried swapping to backup server (identical configuration) to find out if it is not 
caused by defective network card.

> Here it is again for i82544
> ===========
> # nicinfo /dev/io-net/en1
> INTEL 82544 Gigabit (Copper) Ethernet Controller
>  
>   Physical Node ID ........................... 003048 B037CB
>   Current Physical Node ID ................... 003048 B037CB
> ===========
> And here it is once again for e1000
> ===========
> # nicinfo /dev/io-net/en1
> INTEL PRO/1000 Gigabit (Copper) Ethernet Controller
>  
>   Physical Node ID ........................... 003048 B0CC4F
>   Current Physical Node ID ................... 003048 B0CC4F
> ===========

No, you don't need any specific parameters. Qnet logs messages to
sloginfo, so I want to see what it displays. I have been removing and
re-inserting cables in both en0 and en1 without any problems. I have one
link connected with Qnet and the other with TCP/IP and am running tests
on both links.

Hugh.

-----Original Message-----
From: Oleh Derevenko [mailto:community-noreply@qnx.com] 
Sent: Wednesday, May 13, 2009 8:48 AM
To: general-networking
Subject: Re: RE: RE: RE: RE: RE: How to find out if that's a HW fault?

> Please can you get me the sloginfo output after the customer discovers
that the link is down.

Do I need any specific parameters for system to get a good output from
sloginfo? For example, we do not run instrumented kernel - just a simple
one.

> I have been testing here all night without any problems. I have also
been removing cables and re-inserting them while communication is in
progress, without any problems.

EN0 never dies. It's only EN1 on motherboard with two interfaces.

_______________________________________________
General
http://community.qnx.com/sf/go/post29343

> 
> I might take a moment to repeat the suggestion, of
> reconfiguring the PCI bus to get rid of shared interrupts.
> 

I do not have access to that hardware. I live in different country, on the other end of continent.  And can't even check
 if BIOS supports that reconfiguration. With integrated hardware it is not that you just pull a card out and insert it 
into another slot to change interrupt. :(
Let's see if there will be any effect after USBs are disabled and/or io-usb killed.

Hi Oleh,

Sorry, I have been away on vacation. So I guess that this problem is
resolved. Somehow that USB must have been interfering with the Ethernet,
but without having the hardware, I couldn't tell you why.

Hugh.

-----Original Message-----
From: Oleh Derevenko [mailto:community-noreply@qnx.com] 
Sent: Thursday, May 21, 2009 9:35 AM
To: general-networking
Subject: Re: RE: RE: RE: RE: RE: How to find out if that's a HW fault?

Hi, Hugh

Sorry, I don't have anything yet because our guy seems to have made all
the changes I sent him at once instead of trying three different tests.
:) However it looks like the faults do not hapen any more. So, I
suppose, disabling USB helped.

> Please can you get me the sloginfo output after the customer discovers
> that the link is down. I have been testing here all night without any
> problems. I have also been removing cables and re-inserting them while
> communication is in progress, without any problems.
> 
> Thanks, Hugh.
> 
> -----Original Message-----
> From: Oleh Derevenko [mailto:community-noreply@qnx.com] 
> Sent: Tuesday, May 12, 2009 4:49 PM
> To: general-networking
> Subject: Re: RE: RE: RE: RE: How to find out if that's a HW fault?
> 
> > Looking at the PCI interrupts, interrupts 10 and 11 are shared
between
> > the Network and USB controllers. Is the customer running any USB
> > devices? If not, can they slay io-usb and see if that helps
anything.
> 
> I've already told them to try disabling USBs in BIOS if that is
> possible. I'll also pass your suggestion regarding io-usb to them.
> 
> > The disk controller is also on interrupt 11. Is disk performance
> > affected at all?
> 
> I don't know. How can we find that out? We hardly ever use any disk
I/O
> - just for logging (very rarely) and for configuration storage.
> 
> _______________________________________________
> General
> http://community.qnx.com/sf/go/post29315




_______________________________________________
General
http://community.qnx.com/sf/go/post29811