Previous Next Contents

7. Frequently asked questions

These questions and answers are based on my e-mail correspondence on the last version. Thanks to all the people having problems... for making a better documentation :)

7.1 Why isn't the image file gzip'd?

Well since most of the files on the disk already are gzip'd ... like the kernel vmlinuz, the initial ramdisk initrd.gz and add.tgz .. I feel it would be a more pain than gain to gzip 'em.

7.2 Why do you use old binaries?

I compile some of the binaries myself.. most of them are "stolen" from redhat 4.2, the reason I use redhat 4.2 and not 5.0, is the c - library... libc5.so is about 300k ... whilst glibc2 is over 3 megabytes... there is no chance that I cant fit that on one disk.. btw: does anyone know how to compile using libc5 with redhat 5.0?

7.3 Will you have X available?

Well I did an experiment and packed the current XFree86 vga 16 server onto a disk, together with xterm, wm2 and xeyes,.. it worked.. it was slow,.. and I can't see the point... if you have room for an X server... why not mount a directory on the harddrive as /usr/ with an x distribution?

7.4 Why isn't there support for <...>?

Well there are loads and loads of thing you can support in linux but support uses disk space and disk space is sparse... my webspace is bit sparse too... so if you need support for something specific.. you have some alternatives: compile your own kerneland modules for it, make someone else do it.. (I think I've heard rumors about an web-bases kernel compiler?),.. as an last emergency resort... mail me the specifications of what the kernel should include, and maybe I'll do it... compiling kernels take time..

7.5 What about having <...> on the disk?

Well I've implemented a web-server... (used boa for ease of configuration) .. I've used to have lynx on a data disk.. but there are just too many programs to create disks for everything.. so if you really want something on the disk? why not implement it yourself?... I'f you do implement for example an well functioning nfs server, or web server on hal91 .. why not create an .tgz file of it that could be untarred from hal91.ini ... or manually from either hal91, a data disk... or a harddrive?

7.6 When booting it displays 'Boot Failed: change disks and press any key.' and stops.

This is a very familiar problem for me too.. you are using a disk with bad sectors.. and syslinux can't load it..

solution:

Go buy a new disk,.. (I had to even tough I have over 100 old unused ones)

7.7 hal91 is bigger than one disk!

This problem is most probably due to your web-browser!, .. providing files over the http protocol isn't always a good solution, sometimes the browser saves an mime-header in front of the files..

solution:

try using an other browser..

alternative

2: give me an ftp account on some machine, .. the http I've got it on is dead slow anyway

alternative

3: perhaps browsers like it better with a standard filename? hal91.gz ?.. could be worth a try even if compressing it is almost meaningless

7.8 Netscape displays the file instead of downloading it!

Never downloaded things before?.. the correct procedure in netscape i pressing left-shift + left-mousebutton over the link... or you could press the right mousebutton and select save to disk or something.. Using lynx .. press 'd' for download.. Internet Explorer: get netscape, because I don't know but it ought to be easy there too. Alexander Bauer came with the following information:


The download with the Netscape Communicator 4.x dosen't work. 
(i've tested it multiple times). With the Internet Explorer 3.02 i get 
the correct filesize.    

7.9 The keyboard settings are incorrect!

Somehow I got the idea that all Linux users use norwegian keyboard layout... ehh.. no .. . but I distributed an image of my system.. now you just have to inspect /hal91.ini on the floppy disk.. and perhaps replace no-map.gz with dvorak.gz or something.

7.10 Why does initrd.gz still remain the same size after removing files?

This question is perhaps obsolete after seperating add.tgz from initrd.gz but the reason is that deleting a file from initrd just makes the space available,.. you have to zero out that space to make the image compress better.. e.g.


gunzip initrd.gz
mkdir mnt
mount initrd mnt -T ext2 -o loop
rm mnt/bin/joe
dd if=/dev/zero of=mnt/foobar
sync
rm -f mnt/bin/joe
umount mnt
gzip -9 initrd    

The 'sync' flushes the disk cache so linux don't fake the dd if= since the file already is deleted.. you must use maximum compression with gzip to me the kernel unpack it... don't know why must be because of buffer sizes or something.


Previous Next Contents