eth0 not working anymore

[schibbl]

Hi!

My sheevaplug run for 70 days without any problem. But last weekend I could not connect via ssh so I used the JTAG. I found the message

Code:

eth0: link up, 10 Mb/s, half duplex, flow control disabled

spilled all over the console and decided to reboot. It did not change the problem.

Please give me a hint what's wrong and how to fix it.

OS: Debian Squeeze 6.0.1 (Sheeva with Linux 2.6.37.1 Kernel)

Thanks,
schibbl

[schibbl]

Nobody has a clue? So I have to return it to NewIT as a warranty case?

[guidol]

Please give me a hint what's wrong and how to fix it.

Hi,
the same problem (with a GuruPlug) could be read at http://www.newit.co.uk/forum/index.php?topic=677.0
With the Guruplug is seems to be a Heat-issue, but with the Sheevaplug doesnt heat up that much....but
if you got a "older" Sheeva it could be a heat-problem with the power-supply.

Did you cool down your sheeva over night before retrying to use it?

Do you got a network with a higher spedd than 10MBit/Half (I think so)?
Did your restart the othe network components (Router/Switch/PC)?

[NewIT_Marcus]

Nobody has a clue? So I have to return it to NewIT as a warranty case?

The best test that I can suggest for you is a test from U-Boot. See if you can load a kernel from your local network:

See here for a procedure that involves using commands under U-Boot:

Code:

setenv serverip 192.168.1.2
setenv ipaddr 192.168.1.147
tftpboot 0x00800000 uImage
tftpboot 0x01100000 uInitrd

[schibbl]

Hi,

I was testing several methods to boot the plug. First the TFTP method (NewIT Kernel 2.6.32.7), second from NAND (NewIT Kernel 2.6.32.7) and also from SD card (SheevaWL Kernel 2.6.37.1). Booting and using eth0 works, but just for about 24 to 36 hours until the sheevaplug is not reachable via network. Connecting the JTAG shows the console containing the messages described before. Just 12 hours later even JTAG does not connect to a usable console.
In conclusion, there is no dependency to the kernel I load, but I am persuaded of guidols argument there be a thermal problem.

After reading this I noticed that the heat of the upper side (where the NewIT logo is) is quite OK, whilst the lower and the connectors side (where USB, eSata and network is) are burning hot. The whole box is sealed so ventilation does not occur. So this might be a problem by design.

How can I solve this? Is there a new charge without this problem (with less heat generation)?

[NewIT_Marcus]

If eth0 is consistently causing problems, under a variety of boot methods, then that suggests there is a hardware rather than a software problem. It could be a failing PSU; that is a reasonable guess. It is more common for a failing (but not yet failed) PSU to cause problems with USB devices, but I think that we have also seen it affect ethernet.

Jason can provide you details of PSU options, if we are to consider this the root cause (whcih I think is a reasonable supposition at this time).