Red Hat Bugzilla – Bug 438592
abnormal termination of installation - UnicodeDecodeError
Last modified: 2008-03-23 19:59:47 EDT
Description of problem:
Attempted an install via pxeboot using the images available early on 03-22:
initrd.img 21-Mar-2008 04:20 9491648
vmlinuz 21-Mar-2008 04:20 2132448
(timestamp shown from mirror's web directory). Since " Unable to Start X
X was unable to start on your machine", this turned into a text mode
install. It fails with the traceback included below.
Version-Release number of selected component (if applicable):
1 out of 1 attempts
Steps to Reproduce:
1. Boot the pxeboot images, involuntarily select Text Mode install.
2. Within a few minutes of progressing to "Install Starting", with the
progress bar over text such as "6 of 909 packages completed",
rewarded with an exception.
Actual results: Install failed.
Expected results: A successful installation.
Since the text "Save to Disk" appears on screen, but the associated
~checkbox is not selectable, you will have to settle for this manual
transcription (complete with the line breaks as displayed on screen):
-- Exception Occurred --
anaconda 126.96.36.199 exception report
Traceback (most recent call first):
line 160, in callback
s += (hdr['summary'] or "")
it__.py", line 764, in runTransaction
errors = self.ts.run(cb.callback, '')
line 698, in _run
line 682, in run
if self._run(instLog, cb, intf) ==
line 1487, in doInstall
rc = self.ayum.run(self.instLog, cb,
line 209, in doInstall
line 208, in moveStep
rc = stepFunc(self.anaconda)
line 131, in gotoNext
line 701, in run
File "/usr/bin/anaconda", line 1046, in
UnicodeDecodeError: 'ascii' codec can't
decode byte 0xc2 in position 19: ordinal
not in range(128)
OK Save Debug
(In reply to comment #0)
now 2 out of 2 attempts
To slightly improve your chances of recreating this failure,
I thought I should try to gather a few clues for what repository
data was being fetched to install at the time of this traceback.
So I repeated the installation attempt (on the same day, before
any repo updates may render such clues less meaningful).
As an aside, using the boot parameter "xdriver=vesa" did not spare
me the "Unable to Start X X was unable to start on your machine"
circumstance, so this was once again a text mode install.
Watching carefully, I am certain that this traceback occurs at the
time "21 of 905 packages completed" is displayed. No package name is
shown at that time. (I had deselected the "Office and Productivity"
~checkbox on the package selection screen with hopes of a faster
install, and a plan to install additional packages later.) I am
moderately certain that the previous package was so small and
downloaded so quickly that neither "20" nor its name was ever
displayed. I believe the package name "gnome-backgrounds" was on
screen concurrently with "19 of 905 packages completed".
Additionally, through the vagaries of buffering, this time on the
virtual terminal accessible via Alt+F3, following the traceback quoted
above, there was an additional line:
12:34:19 INFO : in run, screen = <snack.SnackScreen instance at 0xb772198c>
It is likely that these remaining informational and warning messages
are not causative for this incident. On the virtual terminal
accessible via Alt+F4, the following messages were displayed (dating
from the installation phase ~10 minutes prior to this traceback
failure, at the time the partitioning information was being written to
<30>Mar 22 12:23:40 hald: 12:23:40.552 [I] osspec.c:288: /proc/mounts
tells, that the mount tree has changed
<30>Mar 22 12:23:40 hald: 12:23:40.570 [I] osspec.c:241: SEQNUM=1115,
ACTION=remove, SUBSYSTEM=vc, DEVPATH=/sys/devices/virtual/vc/vcs5,
<30>Mar 22 12:23:40 hald: 12:23:40.571 [I] device.c:4371: remove_dev:
<28>Mar 22 12:23:40 hald: 12:23:40.571 [W] device.c:4375: Error removing
<30>Mar 22 12:23:40 hald: 12:23:40.583 [I] osspec.c:241: SEQNUM=1116,
ACTION=remove, SUBSYSTEM=vc, DEVPATH=/sys/devices/virtual/vc/vcsa5,
<30>Mar 22 12:23:40 hald: 12:23:40.584 [I] device.c:4371: remove_dev:
<28>Mar 22 12:23:40 hald: 12:23:40.585 [W] device.c:4375: Error removing
*** This bug has been marked as a duplicate of 437993 ***