[coLinux-users] coLinux and/or xp sp3 oddness with MAC address on eth devices

[Jason <ja...@co...>]

I'm posting this as an advisory/hint to anyone having problems with 
0.8.0 and xp sp3. Due to the sequence of events, I don't know if it's 
the 0.8.0 release, or the sp3, or an interaction between them that made 
this happen. Based on a note I found in the readme, I suspect it's a 
behaviour chane in 0.8.0.

I formated my XP system today and brought it back up as XP sp3.

One of the first things I tried to do after installing all teh drivers, 
etc.. was get coLinux working again. Attempting to bring up the 0.7.2 I 
had running before caused immediate reboot.

So I decided to take the next step, figuring sp3 oddness, so I 
downloaded a 0.8.0 snapshot.

I brought up the coLinux from a CMD window without a problem. Everything 
worked great. So I shut it down, installed as a service, and started it 
up again.

eth0 failed to start. I shut it down, brought it up from CMD again, no 
problem. Shut down, start the service. eth0 failed. Looking at it, I 
could see an eth1.

The more astute out there probably know where this is going already.

Looking around a bit I found a hint for Debian to modify the rules that 
udev uses for persistent numbering. Unfortunatly, the debian advice 
didn't work for me (I use Gentoo) and commenting out the contents of the 
file just generated and eth3 and eth4 in my testing.

It was then taht I noticed the MAC address was different on eth0 and 
eth1 in the file, and each MAC was duplicated in eth3 and eth4.

*light over head*

So, it seems that starting up coLinux 0.8.0 on XP SP3 from CMD and from 
service cause different MAC addresses to be generated. I modified the 
rules file so that what was being identified as 'eth1' would be 'eth0' 
(removed the old eth0. Don't know what would happen if two MACs were 
both set to eth0 in the file).

Started the service, and it worked.