[iyonix-support] Re: EtherK not starting up reliably

  • From: Peter Bell <peter@xxxxxxxxxxxxxxxxx>
  • To: iyonix-support@xxxxxxxxxxxxx
  • Date: Mon, 14 Aug 2006 11:49:15 +0100

In message <44CC0F7D.6070300@xxxxxxx>
          James Byrne <jbyrne@xxxxxxx> wrote:

> Peter Bell wrote:
> > My Iyo has been running perfectly well for over three and a half years.
> > 
> > In the last week, the ethernet interface has been failing to start up
> > properly in the morning when I power the machine up - no lights on the
> > switch (a Netgear Gigabit device), and no connectivity (mail fetch or
> > ping).

I've been away for four days, arrived home last evening, had the Iyonix
on for several hours, fetching email etc. without any difficulty.  This
morning I turned the system, it started up normally, but after about an
hour the ethernet connection dropped.

> > I get the following from EKInfo when the system is in this state:

 [Snip]

> > I'm puzzled at the hardware statistics, showing both TX and RX activity,
> > when I appear to have no connectivity.
> > 
> > A *rmreinit etherk gets everything up and running.
> > 
> > Is this a sign of the psu starting to go out of spec?
> 
> Unlikely, the symptom you get in that case is that EtherK fails to 
> detect the Ethernet chip on the motherboard, which is not happening in 
> this case.

Okay - and the PSU is still reading 5.07V.

> 
> > or...
> > 
> > Is it a sign of a problem with the switch?  Although all other
> > ports appear to be working okay, perhaps I should try rebooting that
> > next time.
> 
> Yes, probably a good idea to try resetting the switch.

I've powered the switch down and back up - this makes no difference to
the status of the Iyonix link.  All the other ports came back up as
normal.  I have tried connecting the Iyo to an alternative port, with a
different cable - this makes no difference either.  I have tried
connecting another device to the port usually used by the Iyo, and the
other device is perfectly happy using that port.

All the indications are that, although the Iyo believes that the status
of the EtherK port is good, it is that, in fact, Etherk which is
failing.

> The EtherK output in your mail shows that as far as EtherK is
> concerned everything is OK. The link is up, it's been detected as
> 1000BaseT, and the driver is transmitting and receiving packets with
> no errors.
> 
> As well as resetting the switch, I'd suggest swapping the Ethernet cable 
> in case it's faulty - sometimes it's the simple things that go wrong. If 
> you're convinced that the switch and cable are OK, try "*EKInfo -v" and 
> look at some of the more obscure statistics to see if they give any 
> hints of something amiss - the EtherK user guide file (which should be 
> somewhere on your hard disc) explains what they mean (to an extent).

: *ekinfo -v
: EtherK device driver for Intel PRO/1000 network controllers, version 0.19
: DCI Version 4.05
: Supported network cards detected: 1
: 
: Unit 0 (location: Motherboard)
:   Ethernet address 00:09:95:00:11:00
:   Controller: Intel 82540
:   Bus type: PCI; Speed: 33MHz; Width: 32-bit
:   Media type: Copper
:   Link is up; 1000 Mbps Full Duplex
:   Controller mode: Multicast, reject frames with errors
: 
:   Frame types claimed:
:     Ethernet 0800    (multicast)   handler=(FC2E6610,FB407414)
:     Ethernet 0806    (normal)      handler=(FC2E6610,FB407414)
:     Ethernet 8035    (normal)      handler=(FC2E6610,FB407414)
: 
:   Hardware statistics summary:
:     TX frames 385    : RX frames 290
:     TX bytes  36,625 : RX bytes  44,278
:     TX errors 0      : RX errors 0
: 
:   Driver statistics (only non-zero statistics are reported):
:     Frames received: 290
:     Unwanted frames: 36
:     Maximum transmit descriptors used: 2
: 
:   Hardware configuration:
:     Transmit Descriptors: 256
:     Receive Descriptors: 80
:     Transmit Interrupt Delay: 64
:     Transmit Interrupt Absolute Delay: 64
:     Receive Interrupt Delay: 0
:     Receive Interrupt Absolute Delay: 64
:     Interrupt Throttling Rate: 488
: 
:   Full hardware statistics (only non-zero statistics are reported):
: 
:     Receive counters:
:       Total Octets Received: 44,278
:       Total Packets Received: 290
:       Good Octets Received: 44,278
:       Good Packets Received: 290
:         Packets Received (64 bytes): 70
:         Packets Received (65-127 bytes): 122
:         Packets Received (128-255 bytes): 11
:         Packets Received (255-511 bytes): 87
:       Broadcast Packets Received: 141
:       Multicast Packets Received: 35
: 
:     Transmit counters:
:       Total Octets Transmitted: 36,625
:       Total Packets Transmitted: 385
:       Good Octets Transmitted: 36,625
:       Good Packets Transmitted: 385
:         Packets Transmitted (64 bytes): 126
:         Packets Transmitted (65-127 bytes): 213
:         Packets Transmitted (128-255 bytes): 33
:         Packets Transmitted (256-511 bytes): 13
:       Broadcast Packets Transmitted: 173
: 
:   PHY information:
:     Extended 10BaseT distance: Disabled
:     Polarity correction: Enabled
:     Cable polarity: Normal
:     MDI-X mode: MDI-X
:     Cable length: 0-50 Metres (+/- 20 Metres)
:     Local receiver status: Not OK
:     Remote receiver status: Not OK
:     Idle errors: 510
:     Receive errors: 1

The only values which appear to change over time are the transmit
counters .... about an hour later:

:     Transmit counters:
:       Total Octets Transmitted: 41,876
:       Total Packets Transmitted: 449
:       Good Octets Transmitted: 41,876
:       Good Packets Transmitted: 449
:         Packets Transmitted (64 bytes): 184
:         Packets Transmitted (65-127 bytes): 213
:         Packets Transmitted (128-255 bytes): 36
:         Packets Transmitted (256-511 bytes): 16
:       Broadcast Packets Transmitted: 237


Output from Eklog appears normal

: *eklog 0
: EtherK has logged 4 messages for unit 0.
:   09:31:32 14-Aug-2006: Initialising device
:   09:31:32 14-Aug-2006: Device initialised successfully
:   09:31:43 14-Aug-2006: Link is up 1000 Mbps Full Duplex
:   11:35:03 14-Aug-2006: Performing diagnostic tests


> If that doesn't get you anywhere, try forcing the link speed to 100M
> and see if that helps.

Can this be done on the running system, or is it only via the Configure
eklink command, to take effect on the next start up?

As before, Rmreinit Etherk brings the link back up successfully:

: *rmreinit etherk
: *eklog
: EtherK has logged 3 messages for unit 0.
:   11:41:02 14-Aug-2006: Initialising device
:   11:41:02 14-Aug-2006: Device initialised successfully
:   11:41:04 14-Aug-2006: Link is up 1000 Mbps Full Duplex

Has anyone any clues as to what is going on here?

-- 
Peter Bell - peter@xxxxxxxxxxxxxxxxx
---
To alter your preferences or leave the group, 
visit //www.freelists.org/list/iyonix-support
Other info via //www.freelists.org/webpage/iyonix-support

Other related posts: