[Bug 1692127] Re: duplicate signature for Package problems can be too short

[Brian Murray]

The upload of apport which fixed this bug did not cause but revealed an
autopkgtest failure with test_signal_crashes.py in apport. The tests
test_core_dump_packaged and test_core_dump_unpackaged are failing
because the core file being produced is greater than the ulimit set for
core files. This limit is arbitrary and I've bumped it from 1 million to
10 million bytes to allow the test pass.

-- 
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to apport in Ubuntu.
https://bugs.launchpad.net/bugs/1692127

Title:
  duplicate signature for Package problems can be too short

Status in apport package in Ubuntu:
  Fix Released
Status in apport source package in Xenial:
  Fix Committed
Status in apport source package in Yakkety:
  Fix Committed
Status in apport source package in Zesty:
  Fix Committed
Status in apport source package in Artful:
  Fix Released

Bug description:
  [Impact]
  apport is creating a shorter than appropriate duplicate signature in apport-package bug reports which can lead to misconsolidation and confused developers.

  [Test Case]
  1) Boot a system with an old kernel installed or a freshly installed system
  2) edit /etc/default/grub and replace a ` with a '
  3) Upgrade the kernel, watch it crash
  4) Discover a crash report with a DuplicateSignature whose 2nd line starts with "Examining"

  With the version of the package from -proposed the DuplicateSignature
  in this case should really have a 2nd line that starts with 'Setting
  up linux-image* ...'. While in this particular case not much is lost
  in other cases valuable information will be missing.

  [Regression Potential]
  The code change is correct and has been tested in artful, additionally I'm using the same regular expression to clean up apport-package reports as they come in to LP. The only risk here is a typo in the patch that would cause the hook (the change is in the ubuntu general hook) to error out which apport would handle gracefully and we just wouldn't have a DuplicateSignature.

  [The Original Description]
  Steve brought up bug 1691983 and how it's duplicate signature is missing information.  It contains:

  package:shim-signed:1.27~16.04.1+0.9+1474479173.6c180c6-1ubuntu1
  Installing for x86_64-efi platform.
  Installation finished. No error reported.
  Running in non-interactive mode, doing nothing.
  dpkg: error processing package shim-signed (--configure):
   subprocess installed post-installation script returned error exit status 1

  But is missing:

  Setting up shim-signed (1.27~16.04.1+0.9+1474479173.6c180c6-1ubuntu1) ...
  locale: Cannot set LC_CTYPE to default locale: No such file or directory
  locale: Cannot set LC_MESSAGES to default locale: No such file or directory
  locale: Cannot set LC_ALL to default locale: No such file or directory
  debconf: unable to initialize frontend: Passthrough
  debconf: (Cannot connect to /tmp/aptdaemon-7jrqb_du/debconf.socket: Connection refused at (eval 18) line 3.)
  debconf: falling back to frontend: Noninteractive

  This is because of the following code in data/general-hooks/ubuntu.py:

  145             PKG_MSGS = ('Authenticating', 'De-configuring', 'Examining',
  146                         'Installing ', 'Preparing', 'Processing triggers', 'Purging',
  147                         'Removing', 'Replaced', 'Replacing', 'Setting up',
  148                         'Unpacking', 'Would remove')
  149             for line in termlog.split('\n'):
  150                 if line.startswith(PKG_MSGS):
  151                     dupe_sig = '%s\n' % line
  152                     continue

  The shim-signed package prints a line starts with "Installing " and
  that matches PKG_MSGS so we reset the dupe_sig because that's a
  message we'd expect from a package manager.

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/apport/+bug/1692127/+subscriptions