Re: Cobalt Qube 2

["Rainer M. Canavan" <debian_mips@canavan.de>]

Andy Barnes <andy@barnes.net> wrote:
> I have actually resolved this issue now. I don't know what was causing the
> problem but it seemed the built in firmware was not 'net-booting' or
> something about the configuration steps in Michael's guide are not correct
> for the stock firmware.
> 
> I have moved the drive from my working Qube2 into the non-working one and
> flashed the latest CoLo 1.22 onto the actual device.
> 
> They now both work like a charm. 

Do they? I haven't gotten my Qube2, which has been running 
Lenny (with an old 2.6.22 Kernel) for ages without issues to run 
reliably at all. It basically crashes on the lightest network load
(that can sometimes be recovered with  ifdown eth1; ifup eth1
(or eth0, whichever is the problem). It locks up hard about every
other day or so, and can be made to crash reliably by just scp-ing
large files over the local 100Mbps interface. It's just a router,
nameserver and sometimes-squid proxy on a 6Mbps DSL line. Of course, 
it also crashes basically immediately with the dd/netcat test I've 
posted here in  2010:

> dd if=largefile1 of=/dev/zero &
> dd if=largefile2 of=/dev/zero &
> dd if=largefile3 of=/dev/zero &
> dd if=largefile4 of=/dev/zero &
> tar -cf - largefile* | netcat otherhost 9999

> with the largefiles being a few 100 MB, and a
> netcat -l -p 9999 > /dev/null running on otherhost.

I've tried the stock kernel, a 3.1.6 and 3.2.1 with various settings
for sata/libsata/dma but so far, the tulip driver and settings 
are the same as the stock squeeze kernel. Does anyone here have 
suggestions what I could do to get this thing reliable again?

thanks,

rainer