[sf-lug] debian base system (initially without X11) install progress (or lack there of!) report

Rick Moen rick at linuxmafia.com
Wed Dec 12 00:40:35 PST 2018


Quoting Alex Kleider (akleider at sonic.net):

> I haven't given up!
> Going through the process again I discovered that during the install
> the 'dpkg' command is unrecognized!
> To be clear, what I did was the following:
> When the installer asked for the wifi drivers, I used Alt f2 to get
> to another console from which the following commands were issued:
>  mount -t ext4 /dev/sdb1 /mnt
> cd /mnt
> dpkg -i firmware-.....
> and the system informed me that dpkg was not available!

It's been ages and ages since I last had to deal with this stuff.  My
recollection was that I had no difficulty finding dpkg.  If you think
about it, the installer absolutely needs apt-get and dpkg after
installing the base system, because those are what are required to do
subsequent package operations.

Maybe you didn't got far enough into installation of the base system to
get that and other tools.  Didn't I stress that you will want to go
_back_ in the series of installer screens to the one that recognises and
configures network interfaces?  One of the strengths of Debian's d-i
installer (as with RHEL/CentOS/Fedora's anaconda installer) is that you
can at any point (up to the reboot at the end of installation of the
base system) go back and revisit any of the prior screens.

And, by the way, your phrase 'was not available' is once again your
interpretaion rather than raw data.  I don't want to sound too picky,
but it would be some much better if you'd taken real-time notes and
detail wat you did and what specific responses occurred, in
chronological order with sufficient context.

Probably you mean that at some unstated part of the installer process,
you switched to another console and typed 'dpkg -i [packagename]' as the
root user, and the shell responded 'dpkg: command not found'.  But the
context often matters and the details often matter.  When your helper
has to guess the context and the details, the likelihood of success is
greatly reduced.

After Eric Raymond and I wrote 'How to Ask Question the Smart Way' to
help people more fruitfully get technical help, I kept being amazed at
people continually posting vague and somewhat inaccurate recountings of
what went wrong, what the symptoms and error messages were.  It took me
a long time to figure out the remaining reason why, even after we gave
people clever aids like 'Treat diagnosticians as if they're from
Missouri.  Show them.'    The problem?   I forgot that most people don't
realise your helpers are going to bsolutely need (relevant excerpts
from) accurate contemporaneous records -- including things like the
command you typed (and the context, i.e., at what _point_ in the
installer you did this), and the exact error message the shell gave in
response.

Depending on the problem, your helper may need to Web-search on details
including error messages, and there's a huge difference between 'dpkg
was not available' and 'dpkg: command not found' when Web-searching.

In hindsight, I know what happens:  People do _not_ bother keeping
contemporaneous, accurate notes of the sequence of events when things go
wrong.  They wait until afterwards ds and post their somewhat hazy
post-hoc recollection of events, which inevitably ends up being mostly
their interpretations rather than raw data -- instead of doing the more
time-consuming but wiser thing of going back and doing it again, taking
notes the second time.  So, both the querent and the helper end up being
frustrated.

(At some point, I need to send a patch for the essay to advise readers
that, no, it's really not good enough to just go with what you remember.
If you didn't keep notes the first time through, you really ought to go
back and taken notes during a _second_ run-through before posting your help
request.)


> Is this something worth bringing up in a Debian forum/list?
> 
> Alex



More information about the sf-lug mailing list