Bug 5749 - Cannot upgrade from hard drive
Summary: Cannot upgrade from hard drive
Alias: None
Product: Red Hat Linux
Classification: Retired
Component: installer
Version: 6.1
Hardware: i386
OS: Linux
Target Milestone: ---
Assignee: Jay Turner
QA Contact:
Depends On:
TreeView+ depends on / blocked
Reported: 1999-10-09 03:11 UTC by hchcheng
Modified: 2015-01-07 23:38 UTC (History)
7 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Last Closed: 1999-11-09 17:35:14 UTC

Attachments (Terms of Use)

Description hchcheng 1999-10-09 03:11:52 UTC
I have downloaded all of RedHat directory from an ftp site
to my local harddrive, which is on /dev/hdb1, a Win98
partition.  I am upgrading from RH 6.0.  After the "Reading
package information" dialog, it gives an "Exception
Occurred" dialog (made more concise here):

  Traceback (innermost last)
    /usr/bin/anacond.real     Line 225
    /tmp/lib/python1.5/site-packages/text.py  Line 1000
              ..                                    251
              ..                    /todo.py       1147
              ..                                    942
              ..                                    930
              ..                    /harddrive.py    45
              ..                    /isys.py          8
    System Error: (16, 'Device or resource busy')

The dialog gives two choices "OK" and "Debug".  I chose OK
and everything is umounted and the system is halted.

Comment 1 hchcheng 1999-10-11 19:59:59 UTC
I have just tried putting everything into an ext2 directory
(/dev/hdb3) and install it from there, but the result is exactly the

By the way, hitting "Enter" at the boot: prompt is supposed to start
graphical install, but I get text install instead.  I can't figure out
why.  I think it may have tried to read my CD-ROM drive (from what I
gather in other virtual consoles) and failed.  Does GUI install
require CD-ROM?

Comment 2 pascual_alonso 1999-10-13 17:32:59 UTC
Updating or Installing I'm getting the same error!

------- Additional Comments From   10/14/99 00:48 -------
Yes, I have a brand new P-III dual processor system.  I downloaded
the entire 6.1 onto a clean ext2 HD.  I am going for a brand new
install, not an upgrade.  Everything goes great until "reading
package unformation" is up for a few minutes --- After which, I get
the exact same error!  I have tried various different setup
parameters, all lead to this exact same error message.

------- Additional Comments From   10/14/99 14:58 -------
This is the same as a bug I previously entered. Check out #5672.

------- Additional Comments From   10/15/99 14:20 -------
I was getting a similar error, with the debug I found that a rpm was
incompleted, I got it again and the instalation worked.

The line 45 in harddrive.py returns the list of rpm files as objects,
so if a rpm file has 0 bytes or is incompleted it is included in the
list anyway causing problems later.


Comment 3 Anonymous 1999-10-15 20:27:59 UTC
I have downloaded the entire dist onto ext2 partitions twice now,
from two different mirrors (about 15 hours each!).  I get the same
crash each time.  Running the python debugger, the file isys.py at
line 8 it trys to unmount the disk storing the source distribution
return _isys.unmount(what)

the debugger says the value for 'what' is:'/tmp/hdimage'
From VT #2 I ran the 'mount' command.  I have /dev/hdc1 (my source
tree) mounted on /tmp/hdimage

I attempted to unmount the drive manually, with
a 'umount /tmp/hdimage'.  I got the same error, Device or Resource

My questions are:
1) why is setup trying to unmount the source partition.  It's going
to need those files!!!!
2) Where can I find the source to the boot.img so I can try to figure
out what it's trying to do and debug this myself.
3)  If I were to burn my own CDROM with the source image, would it

I still have not experienced V6.1 and that sucks!  Lets get this

Comment 4 Jay Turner 1999-10-21 12:13:59 UTC
Can someone confirm whether this is indeed because of the GTK.py and
gtk.py files being put on an msdos system (which will see them as the
same thing!)  In other words, is this bug a duplicate of #5672?
Please reopen is it is not a duplicate and see the details in #5672 if
it is a duplicate.

Comment 5 hchcheng 1999-10-21 12:56:59 UTC
This is the same bug as 5672, but it is not a problem with GTK.py vs.
gtk.py (I don't believe 5672 is about that either!).  I have
downloaded everything to an ext2 file system and the result is exactly
the same.

------- Additional Comments From   10/21/99 09:15 -------
5672 had the same symptoms, but was discussing MSDOS file systems.
The problem we note here is upgrading from hard disk. The problem
manifests itself when python attempts to unmount the current hard
disk image while going through mounting/unmounting partitions un the
upgrage package checks. The current hdimage cannot be unmounted
(obviously :^) and the upgrade fails. The python code should be
skipping over the unmounting of the current hdimage.

Comment 6 Jay Turner 1999-10-22 18:58:59 UTC
*** Bug 5828 has been marked as a duplicate of this bug. ***

when I boot from the boot.img (local boot diskette image)
at the scaning for the available packages phase during the
installation, there is an error (exception message), and it
refuses continuing the installation.

Is something wrong in the installable files or there is an
error in the installation process.

Please send me an answer ....

------- Additional Comments From jturner  10/11/99 11:19 -------
Please send the text of the error message to help with this debugging

------- Additional Comments From notting  10/21/99 14:49 -------

*** Bug 6186 has been marked as a duplicate of this bug. ***

I have previuosly sent a bug report on error on installation
of RedHat 6.1 bug # 5828, and no one has answered me so far,
please I need an answer, as I have tried everything I can.

Waiting for your feedback .....

Comment 7 opie.simons 1999-10-22 23:50:59 UTC
From opie.simons:

Glad I found everyone here. I thought our bug was forgotten. :-P

The full text for the Python error message can be found in my
descritpion at the top of Bug #5672. Also to clarify, 5672 was an
install from hard disk and I removed the GTK vs gtk problem by trying
to reinstall from ext2. I see others have also run the test and found
a reproducible problem as well.

jturner- If there is something you'd like me or your dedicated
testers :-) to test or try out, please let me/us know.

Comment 8 hchcheng 1999-10-23 01:21:59 UTC
Just want to add that the newest update image (1999:045) gives the
same result, with slightly a different error message: I think the file
comes from /tmp/updates (or something like that) instead.  The error
is still the same: Device or resource busy.  Let me know if there is
anything you want me to try.

Comment 9 wakeup2 1999-11-03 16:36:59 UTC
I was having the same problem (System Error: (16, 'Device or
  resource busy') and resolved it by comparing the RPMS directory
and found that one of the RPMS wasnt completed.  I resumed the file
and it got past that point fine.  If you're using windows,
get cuteftp and it has a compare directory feature.  I saw
that someone else used this to fix their problem as well.

------- Additional Comments From   11/05/99 16:45 -------
From opie.simons:

I had some time a few minutes ago to check on wakeup2's
comments and though I couldn't use cuteFTP, I downloaded the ls-lr
file from the RPMs dir. on the mirror and compared it to my local
install RPM directory.

I found that the how-to-chinese... file size didn't match. I
downloaded it again, verified the size, and then made it successfully
through the upgrade without changing any other parameters that I have
previously discussed. I did NOT have this RPM previously installed nor
did I upgrade it.

If others have this same problem, the error we are receiving is
definitely misleading since it doesn't point to the real error.

Comment 10 Jay Turner 1999-11-09 17:35:59 UTC
The message is really not that misleading, just pointing at the wrong
place.  What is happening is that the installer was not able to find
the file you were missing and therefore it was falling out.  Well,
since the install source was mounted, the installer was not able to
unmount it and therefore the message about the device being busy.
There is really no way that we can get around this seeing at this is a
python traceback and all it is telling us is what was on the stack at
the time that there were problems.  I am closing this bug, as the
problem is solved.

Note You need to log in before you can comment on or make changes to this bug.