s/where/were/, spotted by nthykier

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

Even if leaving owned files was downgraded to a warning,
there should be a report of these logs.
Do not distinguish between owned-by-one and owned-by-many.

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

Even if leaving unowned files was downgraded to a warning,
there should be a report of these logs.

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

document #316521

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

Add a report for passed packages that leave unowned files in
/etc/alternatives (and probably dangling symlinks in /usr/bin)
after purge.

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

Add a report for failed packages that leave unowned files in
/etc/alternatives (and probably dangling symlinks in /usr/bin)
after purge.

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

In some cases during testing a maintainer script failure is
ignored, e.g. a failed remove that is cleaned up by following
... --pending ... and ... --purge ... commands.
Anyway, there should be a report about these.

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

for packages that Replaces without Breaks another package

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

for failures to upgrade from oldstable to stable to testing

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

for failures to upgrade from oldstable to stable to testing

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

* it is needed for the lenny2squeeze tests, too
* there are more packages depending on lvm2

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

The lenny kernel needs some preseeding to allow installation
without bootloader in a chroot ...

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

Allow to remove the kernel image inside the chroot even it its
version matches the kernel running on the host.

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

sync detect_archive_issues and dependency_error.conf

apt in experimental now reports
  E: Unable to correct problems, you have held broken packages.

update the description

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

If the tempdir runs full, tar will fail to extract the basetgz.

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

Since the slave may deliver new logfiles while the daily
processing runs
  detect_piuparts_issues
  detect_network_issues
  report_newly_bugged_packages
there is a possibility for logs with piuparts or network issues
to be classified as bugged by piuparts-analyze.

Just run the detection of network and piuparts issues for
bugged/*.log, too

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

Add 'set -x' in the postrm scripts of $PIUPARTS_OBJECTS before
running 'postrm purge'.

This is helpful for packages failing in the purge phase to find
the actual failing command.

This is only intended for debugging purge failures and should not
be enabled by default.  But you can easily rerun the piuparts
command manually and add another instance of --scriptsdir to
include this script.

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

a slight modification of the unowned_files_after_purge template

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

For reporting uninstallable packages that definitively need a fix
in that package, i.e. a problem not caused by ongoing transitions
and not fixable by a plain binNMU.

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

for sid->experimental file overwrite errors

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

having (, ), | in filenames complicates using them in the shell
as these chars need escaping

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

The mirror may be updated during piuparts tests.  Depending on
the actual timing, this leads either to errors fetching or
validating files (already covered by other checks) but in some
rarer cases may cause some successful package updates that were
not done in the reference snapshot.
This usually leads to modified changelogs, so let's check for
them and reschedule the failed package.
I can't imagine a buggy package causing modified changelogs
(except for perhaps a postinst deliberately recompressing
foreign changelogs with different options).

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

If the mirror is not reachable while fetching .debs, we get
  Failed to fetch.*Unable to connect to.*

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

      <b>2012-02-20</b>: piuparts-analyse now sends commands the BTS: if a bug has not been explicitly marked fixed in the new version, it can rather very savely be assumed it's still present.

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

taken from http://debian.semistable.com/debgraph.out.html



Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

for failing a install-remove-distupgrade install test,
i.e. install from config-files-remaining (from previous release) state

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

for failing a install-remove-install test, i.e. install from
config-files-remaining state

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>

before they reach testing.

with pointers to ucf instructions and examples

Signed-off-by: Andreas Beckmann <debian@abeckmann.de>