[PLUG] HELP! Broken Ubuntu! Can't login!

Keith Lofstrom keithl at kl-ic.com
Sun Dec 11 01:18:26 UTC 2005


On 10 Dec 2005, at 15:45, Keith Lofstrom wrote:
> First, in the there have been problems with USB external drives and
> Linux.  I found a chipset dependence, but did not have time to pursue
> it.  At the time, with both late 2.4.XX kernels and early 2.6.X
> kernels, the process would go for a while and then hang.  I do not
> know what the status of the problem is now.  If it still is a problem,
> then some careful attention and tracing would probably provide enough
> detail that the USB driver folks could take it from there.  I suggest
> loaning the drive to somebody else, and see if they see similar
> problems.  That's why we're here, guy!
 
On Sat, Dec 10, 2005 at 03:58:17PM -0800, John Jordan wrote:
> The Live CD sees the external USB drive just fine. The archive file 
> made with Dump two days ago is still there -- 9 Gb. And Dump had 
> no problem writing to this USB drive. So I seriously doubt there is a 
> hardware or driver problem with the new USB drive.

I was unclear.  The problem is low rep-rate and hard to reproduce - I
would see it about once per 10GB moved, sometimes after a few hundred
megabytes and sometimes after 50GB.  It seems to relate to
synchronization and error recovery.  You can find a writeup at
http://www.keithl.com/usb2bug.html , and mentions of it in the PLUG
archives and elsewhere.  If you have repeatedly filled and emptied your
external drive without error, then you can rule out hardware errors.


> plastic wrappings. I don't think this is a hardware problem.
> 
> In fact, other than flaky software, I have had no problems at all with 
> this computer. This problem was created by Kdar. No question 
> about it. 

But without the actual bug detected or at least characterized, how can
you tell?  You may have no question about it, but the people you expect
to solve the problem for you do have questions.  Questions are how 
answers are found.   You have learned that you cannot make Kdar work
for you using the thought and skills you are willing to apply.  That
does not necessarily mean the code is broken, but it does mean it is
not suited for you.

As a person that designs hardware, let me assure you that none of it
is perfect.  "New in the plastic wrap" means somebody knows how to
put a unit in a bag and seal it in a box.  So do the clowns at Fry's,
and they sell "defective new in the box" stuff all the time.  Even
the best hardware from unimpeachable sources comes with failure
rates approaching a percent.  Sometimes every instance of a product
is defective, and the manufacturer sells the dreck anyway.

I suspect software is indeed the problem, but the process of fault
isolation requires slicing the problem into pieces, and the 
hardware/software slice is usually good for cutting a problem in
half.  However, intermittent software problems that behave like
you are describing typically have their roots in infrequent conditions
in hardware, which should be detected and bypassed and are not.
Since the problem is locking up your machine, it is likely something
down in the kernel or drivers or below - it is very unlikely that
any user space code, even code run by root, can cause a machine to
lock up without help from the kernel and the hardware.

> I think that the best solution right now is to use that Dump archive 
> file to restore. Unfortunately, I have no idea how to do that. I 
> googled and found some instructions for Dump/Restore, but they 
> were written in Geekian. I understood one or two words per 
> sentence, usually things like "the" and "an." :(

And geekian - or books, or something else that requires an investment
of effort, is how you are going to find a solution.  Dump is not as
hard as most things in Linux, and it has been around long enough that
you can find instructions in 30 year old books.  I can dump my usr
partition with something like  "dump -0f dumpfile /usr" - that is 
about as easy as it gets.  File maintenance, and restores, require
quite a bit more effort, which is one reason why I do not like the
apparent simplicity of dump.

I no longer use dump, and maintain an open source disk-to-disk backup
program, dirvish.  However, dirvish is a lot more complicated and 
powerful than dump, so I would not suggest it in your case.  It does
mean I have seen a LOT of disk-related and USB-related errors from
dirvish users, which is why I believe that intermittent hardware 
failure, or kernel code that does not bypass those failures, is a
distinct possibility in your case.

> Failing that, I could just reinstall from the Ubuntu-64 Install disk. 
> But that will wipe out everything, I think. Lots of work to put 
> everything back. I tried the Install disk just now, but it doesn't seem 
> to have a Repair option.

A lot of the questions you ask are answered by some good books.  I 
suggest Carla's "Linux Cookbook", Peter Harrison's "Linux Quick Fix
Guide", and the "Linux Administration Handbook" by Nemeth.  These are
not self-licking ice cream cones, however;  the sweetness will require
some effort to get access to.

Linux for the desktop is still arriving.  It has not become effort-free
to use yet, though it is getting closer daily.  If you want something
closer to effort-free, you may want to consider a Mac.  Linux is (IMHO)
easier to use than Windows, in the sense that the sum of easy and hard
things necessary to safely operate a computer are easier as a whole for
Linux (try doing a bare-metal disk restore from backups in Windows XP). 
But for now Macs are easier still, and allow you to substitute money
for effort.  If you can't afford either money or effort, then you will
have many disappointments.

Keith

-- 
Keith Lofstrom          keithl at keithl.com         Voice (503)-520-1993
KLIC --- Keith Lofstrom Integrated Circuits --- "Your Ideas in Silicon"
Design Contracting in Bipolar and CMOS - Analog, Digital, and Scan ICs



More information about the PLUG mailing list