Infrequently Asked Questions Edit
Problems experienced getting provided Debian-3.0r0.ext3.1gb to work Edit
Q1. How do you get DHCP working with coLinux Debian-3.0r0.ext3.1gb image?Edit
dhclient fails with following log messages:
Feb 26 13:22:01 colinux dhclient-2.2.x: socket: Protocol not available - make sure CONFIG_PACKET and CONFIG_FILTER are defined in your kernel configuration! Feb 26 13:22:01 colinux dhclient-2.2.x: exiting.
I haven't had any problems like this with DHCP using coLinux ver. 0.5.4. The kernel should already be configured properly to allow DHCP. However, I think I added pump (DHCP config client) to the Debian image first because I knew I needed it. If the command 'dpkg -l pump' indicates it's not installed, first configure eth0 manually with a static IP address. Then do an 'apt-get update; apt-get install pump' to install it. Then reboot to see if that fixes your problem.
rak (email@example.com) : That's great for people who can access the internet with a static IP, but it would be cool if someone could suggest a way of installing a DHCP client that doesn't involve already having a working network connection. I'm not sure why the base image maintainer hasn't seen fit to add a DHCP client...
mju: Using Debian-3.0r2.ext3-mit-backports.1gb.bz2 (Size: 18329320) "pump" is already in the image. Booting coLinux the first time gave me a fixed IP address, using pump -i eth0 (TAPped to my real LAN with existing DHCP service) took a correct IP address. For a permanent DHCP client configuration you have to change the interfaces configuration:
colinux:/etc/network# cat interfaces # /etc/network/interfaces -- configuration file for ifup(8), ifdown(8) # The loopback interface auto lo iface lo inet loopback # The first network card - this entry was created during the Debian installation # (network, broadcast and gateway are optional) auto eth0 iface eth0 inet static address 192.168.0.40 netmask 255.255.255.0 gateway 192.168.0.1
change the last 4 lines to the following 6:
### DHCP client configuration (needs dhclient, pump, udhcpc or dhcpcd): iface eth0 inet dhcp #iface eth0 inet static # address 192.168.0.40 # netmask 255.255.255.0 # gateway 192.168.0.1
The current Debian image (Debian-3.0r2.ext3-mit-backports.1gb.bz2, sized 18329320 bytes) should then use an offered DHCP service if its eth0 is bridged to the managed (W)LAN.
to rak: Where's the problem? Using DHCP means you have a LAN environment. Choosing one time a special IP address manually isn't a hassle, is it!? If needed configure the device in
/etc/network/interfaces followed by the command
Q2. How do you install packages with apt-get?Edit
If you want to have all the useful packages including less, ftp, ssh... you can do an :
apt-get update apt-get install base-config
If it don't start or if you want to do again the base-config, simply launch
In this way, you will be able to choose the timezone, the root password, to add an user, ton configure apt, and launch dselect that will suggest you a set of useful packages like less, ftp...
Once your Debian system is set up, be sure to periodically (daily :^) run 'apt-get update; apt-get -u upgrade; apt-get clean' to keep it updated with the latest security patches, etc.
Q3. How do you use apt-get behind a proxy?Edit
for ftp use ftp_proxy instead, for the rest the syntax is the same.
or (to get through a Microsoft ISA proxy server) :
or alternatively, to use it each time you run apt-get, add the following to your /etc/apt/apt.conf file :
Q4. I installed vncserver but it can't find the fonts it needs?Edit
The VNC Log file says:
colinux:# cat .vnc/colinux:1.log 20/02/04 07:29:03 Xvnc version 3.3.3r2+tight1.2.2 20/02/04 07:29:03 Copyright (C) AT&T Laboratories Cambridge. 20/02/04 07:29:03 All Rights Reserved. 20/02/04 07:29:03 See http://www.uk.research.att.com/vnc for information on VNC 20/02/04 07:29:03 Desktop name 'X' (colinux:1) 20/02/04 07:29:03 Protocol version supported 3.3 20/02/04 07:29:03 Listening for VNC connections on TCP port 5901 Fatal server error: could not open default font 'fixed'
The process complaining about not finding the font 'fixed' is the X server which vncserver runs, not vncserver itself. I think the solution is hinted at on this webpage:
This mentions that to fire up VNC and get a login screen, you need to use the option '-fp unix/:7100' to Xvnc (which is passed on to the X server). The argument 'unix/:7100' means the fonts can be accessed on port 7100 (on the local machine using unix sockets). This requires the X font server (xfs) package to be installed, but it isn't on the base Debian disk image. Probably the easiest way to get it working is:
- Make sure you installed the recommended package xfonts-base when you installed vncserver. It may not be automatically installed for you. Check with 'dpkg -l xfonts-base' to see if the package is found and is currently installed. If not, do 'apt-get update' then 'apt-get install xfonts-base'. The font 'fixed' (actually an alias of one type of 6x13 font) is included in xfonts-base.
- After installing xfonts-base, try starting vncserver again.
- If you still have the same error, add the argument '-fp /usr/X11R6/lib/X11/fonts/misc' at the end of the vncserver command. The misc subdirectory is where the font used for 'fixed' lives.
- If there's still a font problem, install the package xutils and run 'update-fonts-dir misc' as root.
- If that still doesn't fix it, you might try 'apt-get --reinstall install xfonts-base'.
- If you want to try out the X font server, install the package xfs. You may also want to install xfonts-75dpi (or -100dpi) and xfonts-scalable. Then start xfs with '/etc/init.d/xfs start' and try running vncserver with the argument '-fp unix/:7100'.
- (Are there any other things to try? X experts welcome :^)
You might want to look at the pages linked above and this one:
for some neat ideas on how to get graphical access to your coLinux guest from the Windows host.
Q5. Why does my Debian image have fsck problems after an apt-get update?Edit
This has to do with the init script that checks filesystems getting updated. Check the answer to Q10 in the FAQ.
Q6. Run coLinux with Knoppix?Edit
You can't simply boot Knoppix from the cd-image with the colinux kernel, because colinux does not support the compression used on the knoppix filesystem (cloop.o, compressed loop device). The only way to make booting the cd image work directly is recompiling the colinux kernel with the cloop patch applied. However, what does work is booting knoppix from cd without colinux, then installing it to your harddrive by typing "sudo knoppix-installer". Make sure your installation is not a compressed installation. The knoppix partition on your harddrive can now be included in your colinux configuration file. See WindowsDevicePaths for how to add a partition to the Colinux configuration file. Now knoppix can be adapted to work with colinux as you would adapt any other distribution.
Q7. I want to run some older version of coLinux but now I can't !Edit
most coLinux users just intall to directory c:\colinux and work from there. on each release, they keep using the same directory, only the binaries are updated by the installer.
now some other users put their coLinux images in c:\colinux and installed the executable files in other directories (like c:\coLinux0255, c:\co Linux0417 ...), in order to keep old versions of coLinux installed on their disk.
this scheme worked for a while - at least until snapshot coLinux-20040417.exe - but now, the location of linux.sys is put in the registry by the installer, and colinux-daemon.exe relies on this info to launch correctly. (indeed, there is some dependancy between linux.sys and the .exe files).
therefore, if for example version 0509 is installed, you can launch only this version now.
so to remove this limitation, you have to go the later version directory then :
- first unstamp Windows registry like this : colinux-daemon.exe --remove-driver
driver: stopping driver service driver: removing driver service
- then if the version of coLinux you want to run use the registry, go to the other directory and stamp the registry again from there : colinux-daemon.exe --install-driver
daemon: loading C:\coLinux0429\linux.sys daemon: driver installed
now the older version of coLinux should be working again.
Also, if you are using TAP (the TAP-Win32 Adapter), you need to change your line :
<network index="0" type="tap" name="TAP"></network>
<network index="0" type="tap"></network>
that is, remove the "name" item. otherwise, the network will not be functionnal (aka, the TAP icon will stay and show as 'disconnected' in the systray).
Q8. Everything is fine but coLinux gives a GPF at startupEdit
among many other things, there has been noted a tendancy of some dead-brain text editors to silenty switch the content of text files to Unicode (UTF-8) by inserting two bytes at the beginning of the file. of course, the same stupid editors will NOT tell you and NOT show you those two bytes.
the main culprit is Notepad being used to edit the colinux XML configuration file, if one also use special letters (like é or ö) in the same file.
fix : remove the offending two bytes and use a real text editor instead - one you can trust.
Q9. I'm getting an error message "Codepage 437 not found" when mounting a FAT32 partitionEdit
It seems that the mount program first tries to mount partitions of type "vfat" (that the driver for fat32) with codepage 437 instead of 8859. There should be a mount option to change this but I didn't succeed with "-o codepage=8859-15". My solution was to recompile the kernel for my version 0.6.3.
- If you're using 0.6.3: Simply replace your vmlinux file with this_file and everything should be fine.
- If you're using an older version: Upgrade.
- If you're using a newer version: Follow the instructions in this wiki's FAQ and recompile your kernel with the options Filesystems - NLS support - <enable your codepage here by pressing 'Y'>.
Background: As filenames can be coded with different codepages (imagine a chinese partition coded with UTF filenames vs. an age old US fat partitions with pure ASCII filenames, every mount command needs at least some kind of codepage.
Q10. coLinux Boots okay, but my TAP-Win32 is not coming upEdit
coLinux seemed to have trouble in my installation enumerating through the registry to find the first vailid TAP entry. If your TAP is not coming up you can confirm this problem by using colinux-net-daemon.exe, and it will complain that no valid TAP adaptors can be found, even if you have one staring at you in the system tray.
The network adaptors can be found in the following registry key, and I fixed my problem by making the TAP connection the first entry in the list (I simply deleted an empty connection key that was ahead of the TAP connection)
Take care in deleting anything from your Win32 registry - if the connection key is empty it does not necessarily mean hat it is not used, it may actually be a placeholder for a removable device.
MassTranslated on Sun Apr 23 17:35:47 UTC 2006