Reduced bug reports may by actually larger than the full ones

Bug #92653 reported by Michael Hofmann
52
This bug affects 6 people
Affects Status Importance Assigned to Milestone
apport (Ubuntu)
Fix Released
Medium
Martin Pitt

Bug Description

Binary package hint: apport

See attached dump.

Revision history for this message
Michael Hofmann (mh21) wrote :

Complete report (recommended; 1.6 MB)
Reduced report (slow Internet connection; 2.5 MB)

Revision history for this message
Martin Pitt (pitti) wrote :

Reproduced with your report, thank you!

Changed in apport:
assignee: nobody → pitti
importance: Undecided → Medium
status: Unconfirmed → In Progress
Revision history for this message
Martin Pitt (pitti) wrote :

Heh, that's indeed a funny one. gdb produces a huuuuge stack trace on this core dump, so that the reduced report would indeed have been 2.5 MB. The size calculation of the complete report is not 100% correct, though (for various reasons). Fixed in bzr head, which caps the size of the traces.

Changed in apport:
status: In Progress → Fix Committed
Revision history for this message
Martin Pitt (pitti) wrote :
Download full text (5.2 KiB)

 apport (0.70) feisty; urgency=low
 .
   [ Martin Pitt ]
   * bin/apport-retrace: Add option --remove-tag to remove a Launchpad bug
     tag. This is intended for an automatic Malone crash retracing system.
   * debian/control: Bump python-launchpad-bugs dependency to ensure that we
     have Bug.[gs]et_metadata().
   * man/apport-retrace.1: Add documentation for --confirm and --remove-tag.
   * bin/apport-chroot: Add option --remove-tag and pass it to apport-retrace.
   * apport/chroot.py, fix_symlinks(): Convert chroot path prefixed absolute
     symlinks to relative symlinks to avoid fakechroot's weird handling of
     absolute symlinks.
   * Add bin/launchpad-crash-digger: Daemon for watching out for
     need-$ARCH-retrace tagged Ubuntu bugs in Launchpad and calling
     apport-retrace on them.
   * bin/apport-retrace: Mangle bug comment with StacktraceTop to not contain
     invalid UTF-8, to avoid getting Internal Server Errors from LP.
   * debian/local/setup-apport-retracer: Install libc6-i686{,-dbgsym} into an
     x86 chroot, to get sane x86 backtraces for crashes in libc.
   * debian/local/setup-apport-retracer:
     - Unpack and install python-launchpad-bugs locally if the package is not
       installed.
     - Link launchpad-crash-digger into the retracer's bin/ dir.
   * run-tests: Run tests with python's -tt flag to catch whitespace errors.
   * Replace tabs with spaces in all Python files. (LP: #93561)
   * Remove trailing white space in all Python files.
   * apport/report.py, add_proc_info(): Do not regard symlinks to executables
     as interpreted scripts any more (such as Debian alternatives). Add test
     case. (LP: #94732)
   * problem_report.py: Add new method get_new() which returns a set of all
     keys which have been added since load() or construction. Add test cases.
   * problem_report.py: Add optional parameter only_new to write(), which
     writes only the get_new() keys. Add test case.
   * apport/ui.py: Remember currently processed report file and update it with
     the added information, so that it becomes useful for local evaluation,
     too. Bump python-problem-report dependency to ensure write()'s only_new
     availability. (LP: #94678)
   * apport-chroot: Add forgotten sys.exit(1) after printing the error message
     about an invalid chroot specification.
   * apport/ui.py, run_crash(): Check for a field "UnsupportableReason: <text>"
     and display an information box that the current configuration cannot be
     supported because of <text>, instead of processing and reporting the
     crash. Add test case for this workflow. With special regards to our
     Firefox crash triagers who want to get rid of the hundreds of
     flash-related crashes. :)
   * apport/report.py, add_hooks_info(): Use execfile() instead of
     __import__(), since package names might conflict with module names already
     imported into apport's namespace. Also search for hook named after the
     source package name (prefixed with 'source_'). Add test cases.
   * bin/apport-chroot: When specifying --save for login, only save the tarball
     if the exit status is 0.
  ...

Read more...

Changed in apport:
status: Fix Committed → Fix Released
Revision history for this message
Kenneth Belitzky (httpdss) wrote :

im having this same feeling. check screenshot.

ubuntu: 8.10
apport version: 0.116

Revision history for this message
LimCore (limcore) wrote :

Occures for me (xorg crash) on up to date 8.10

Revision history for this message
LimCore (limcore) wrote :

...for me it was because /var/log/Xorg.0.log.old file was 150 MB in size after my intel X crashed ( bug#85298 )

*** another bug: Also, this indicates that Xorg.0.log.old was probably NOT included in the recommended/full crash info

btw, while uploading said crash dump, the upload failed,

Revision history for this message
alm (alm) wrote :

Just encountered this bug on an up-to-date jaunty system (apport 0.145). Apport report a SIGSEV in syndaemon (it actually crashed an hour ago just during shutdown). Complete report was 44.5 KiB, reduced report 458.1 KiB (see screenshot). Submitting the report failed because I didn't have a network connection at that time. I don't know if the submitted information was actually 44.5KiB since it didn't offer a retry option after it failed to connect.

Touching the file in /var/crash triggered apport to pop-up again, this time the correct sizes were displayed (slightly above 500KiB for the complete report).

Revision history for this message
alm (alm) wrote :
Revision history for this message
Martin Pitt (pitti) wrote :

Reopening then.

Changed in apport (Ubuntu):
status: Fix Released → Triaged
Revision history for this message
Elias K Gardner (zorkerz) wrote :

This just happened to me in Karmic.

Revision history for this message
John Lenton (chipaca) wrote :

Yep. I'm attaching the screenshot, because it was lots of fun.

Revision history for this message
Martin Pitt (pitti) wrote :

trunk r1544

Changed in apport (Ubuntu):
status: Triaged → Fix Committed
Revision history for this message
Launchpad Janitor (janitor) wrote :
Download full text (3.3 KiB)

This bug was fixed in the package apport - 1.8.1-0ubuntu1

---------------
apport (1.8.1-0ubuntu1) karmic; urgency=low

  * New upstream bug fix release:
    - data/general-hooks/generic.py: Check $HOME, not /home for enough space.
      (LP: #422658)
    - launchpad.py: Intercept httplib2.ServerNotFoundError as well, to avoid
      crashes when being offline. (LP: #396276)
    - apport-cli: Save reports with .apport extension instead of .txt. Thanks
      to Steve Beattie! (LP: #401983)
    - fileutils.py, likely_packaged(): Ignored crashes in /var, packages don't
      ship executables there, and it creates false positives. (LP: #414368)
    - packaging-apt-dpkg.py, get_modified_files(): Fix crash with empty lines.
      (LP: #408280)
    - packaging-apt-dpkg.py: Use installed version instead of candidate
      version where appropriate. This also fixes a crash where an obsolete
      package is not available any more. (LP: #423511)
    - hookutils.py, attach_gconf(): Fix crash with keys which do not have a
      schema default. (LP: #422277)
    - launchpad.py: Remove LP #353805 workaround, seems fixed now.
    - launchpad.py: Talk to staging if $APPORT_STAGING is set.
    - launchpad.py: Explicitly supply content_type for addAttachment, current
      wadllib requires it now.
    - apport_python_hook.py: Paper over inexplicable import error.
      (LP: #348250)
    - apport_python_hook.py: Protect against nonexisting sys.argv.
      (LP: #418051)
    - apport/ui.py, load_report(): Check that report has ProblemType field.
      (LP: #198543)
    - ui.py: Fix handling of complete vs. reduced report size. (LP: #92653).
      This also fixes a race condition crash with os.path.getsize().
      (LP: #348137)
    - fi.po: Fix mistranslation of "&Cancel". (LP: #355303)
    - apport-{gtk,kde}: Check for having $DISPLAY at startup to avoid crashes.
      (LP: #411276)
    - report.py, add_gdb_info(): Fix race condition in unlink_core, thanks to
      Tommi Komulainen! (LP: #397945)
    - ui.py, load_report(): Robustify check whether program is still
      installed. (LP: #329184)
    - packaging-apt-dpkg.py, install_retracing_packages(): Install package for
      ExecutablePath/InterpreterPath if missing; this can happen with package
      hooks which reassing package
    - launchpad.py: Add a comment when marking a bug as a duplicate.
      (LP: #418871)
  * Move gdb dependency from apport to GUI packages to avoid pulling in gdb on
    Ubuntu server. Thanks to Steve Beattie! (LP: #354172)
  * ubuntu-bug: Fix handling of .crash file arguments, thanks to Marco
    Rodrigues for pointing this out! (LP: #422881)
  * debian/local/apport-collect: Set content_type and description, wadllib
    requires them now. (LP: #423512) Also drop the ASCII reencoding
    workaround, this doesn't seem to be necessary any more.
  * apport/hookutils.py, attach_conffiles(): Fix crash with obsolete
    conffiles. (LP: #412132)
  * debian/local/apport-collect: Do not upload data if the affected package
    isn't installed and there is no source package hook available either.
    (LP: #417277)
  * debian/local/ubuntu-bug: Accept .apport extension, too; thanks to Steve
    ...

Read more...

Changed in apport (Ubuntu):
status: Fix Committed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.