Re: Colo 1.14 Boot Problems / NFS 'udp: no matching socket' fix
On Fri, Nov 19, 2004 at 02:27:41AM -0000, Nick Bartle wrote:
>
> I have recently installed Debian on my Cobalt Qube 2 and have / had a few
> teething troubles:
>
>
>
> When NETbooting:
>
> 'udp: no matching socket'
>
>
>
> The fix that worked for me was editing '/etc/services'
>
> on my nfs server (debian sarge on i386) and adding:
>
>
>
> 'mount' 768/udp'
>
>
>
> It would appear that the Qube's nfs client tries to connect to the NFS
> server via a random udp port < 1024,
>
> whereas the NFS server expects connections portmapped > 1024. With this
> setting in my 'services' file
>
I don't know about this one. CoLo always uses a source port in the range
768 --> 1023 as apparently some servers only accept connections from the
reserved port range. It then connects to portmap to get the port
addresses of the required services. All pretty standard stuff.
>
>
> My present problem is with Colo 1.14. With the default 1.10 installed the
> Qube boots fine; I upgraded to the
>
> New 1.14 via martins .deb package and / or the tar.gz from Peters page and
> the Qube refuses to boot.
>
> It displaysthis on the LCD:
>
>
>
> ! UNIT BOOT !
>
> ! FAILED !
>
>
>
> And shows this in the console (Trimmed off uneeded bits):
>
>
>
>
> 1> lcd 'Booting...' /dev/{mounted-volume}
>
> 1> -load /boot/default.colo
>
> ext2: {boot/default.colo} --> {./default.colo}
>
> 00000093 147t
>
> 1> -script
>
> 2> mount hda1
>
> ext2: revision 0
>
> 2> load /vmlinux
>
> ext2: {vmlinux} --> {vmlinux-2.4.27-r5k-cobalt}
>
> 002a5bfe 2776062t
>
> 2> lcd "Booting Debian"
>
> 2> execute root=/dev/hda2 console=ttyS0,{console-speed}
>
> no such variable
>
This is the problem. The variable "console-speed" is not defined and
CoLo falls back to loading /boot/vmlinux.gz which is of course CoLo ...
bang!
What output does CoLo give you for the following commands
variable
nvflags
noop {console-speed}
?
P.
Reply to: