Installation of application from the click store appears not to work

Bug #1385656 reported by John McAleely
12
This bug affects 1 person
Affects Status Importance Assigned to Milestone
packagekit (Ubuntu RTM)
Fix Released
Critical
Unassigned

Bug Description

Using ubuntu-touch/ubuntu-rtm/14.09-proposed # 126, krillin

From the Ubuntu home screen, select the ubuntu store. Browse to an application you don't have installed.

Select 'Install'

Observe the progress bar go from 0% to 100%, with no further progress after that (button sits at 100%)

Return to the app home screen to see if the app is installed anyway

Actual result:

Observe the app is not installed

Expected result:

The application is installed.

Tags: rtm14
Revision history for this message
John McAleely (john.mcaleely) wrote :

Anecdotal reports from email indicated "I can't even install with pkcon"

Changed in ubuntu:
importance: Undecided → Critical
tags: added: rtm14
Changed in ubuntu:
assignee: nobody → Pat McGowan (pat-mcgowan)
Revision history for this message
John McAleely (john.mcaleely) wrote :
Revision history for this message
Oliver Grawert (ogra) wrote :

i see the following in /var/log/syslog when trying to install a package:

Oct 25 09:17:08 ubuntu-phablet dbus[735]: [system] Activating service name='org.freedesktop.systemd1' (using servicehelper)
Oct 25 09:17:08 ubuntu-phablet dbus[735]: [system] Successfully activated service 'org.freedesktop.systemd1'
Oct 25 09:17:14 ubuntu-phablet dbus[735]: [system] Failed to activate service 'org.freedesktop.PackageKit': timed out

Revision history for this message
John McAleely (john.mcaleely) wrote :

Talking to Pat, we wondered if it was the unity-scope-click landing here:

http://people.canonical.com/~ogra/touch-image-stats/rtm/124.changes

(No proof other than a suspicious sounding name, of course)

Revision history for this message
Ricardo Salveti (rsalveti) wrote :

Downgrading apt fixed the issue for me (it got updated on 125 for security fixes).

Revision history for this message
Ricardo Salveti (rsalveti) wrote :
Revision history for this message
Colin Watson (cjwatson) wrote :

The correct treatment for this is to notice that it's a recurrence of https://bugs.launchpad.net/ubuntu/+source/packagekit/+bug/1368246 due to ubuntu-rtm/14.09 being stale; syncing the fixed packagekit into ubuntu-rtm/14.09 (and un-reverting apt) will fix it.

affects: ubuntu → packagekit (Ubuntu RTM)
Revision history for this message
John McAleely (john.mcaleely) wrote :

That certainly looks like the same bug from the repro steps.

Revision history for this message
John McAleely (john.mcaleely) wrote :

So, for the purpose of recording on this bug, I'll note that it is planned to respin an image (128) with the apt package reverted.

As cjwatson notes, there is what looks like a better fix that should be used for future images.

Revision history for this message
Colin Watson (cjwatson) wrote :

Right, reverting does the job for now, but we should generally regard reverting security fixes as something to be strenuously avoided so we shouldn't consider the matter closed with that.

My suggestion is:

 * binary-sync new packagekit from utopic into 14.09, via a silo if you want
 * get a member of ~ubuntu-archive to un-revert the apt change as follows:
  * remove-package -A ubuntu-rtm -s 14.09 -m 'reverting to 1.0.9.2ubuntu2' apt
  * copy-package --from=ubuntu-rtm -s 14.09 -e 1.0.9.2ubuntu2 -b --force-same-destination apt

(Note that the --force-same-destination option requires lp:ubuntu-archive-tools >= r909, so the archive admin doing this should take care to update their local checkout first.)

apt must be done strictly after packagekit, but it doesn't particularly matter how much later. However, the remove-package and copy-package operations must be done as close as possible to each other, ideally in a single publisher run. The derived distributions publisher runs at "*/5 * * * *" in crontab(5) syntax, so it should be sufficient to run it at a minute ending in 1 or 6 from a system with reasonably reliable networking. The consequences are not permanent if you get this wrong, but it would cause the ubuntu-rtm archive to be essentially unusable for any kind of build operation for a few minutes, so it's best to make some effort to get it right.

I'm on holiday until Wednesday now and don't especially want to be coordinating this from a hotel room over the weekend. Other members of the archive admin team should be able to deal with this given the above description, though, so I suggest grabbing one of them on Monday or Tuesday.

Revision history for this message
Marc Deslauriers (mdeslaur) wrote :

FYI, I have now prepared a backported minimal security patch for apt.
I plan on requesting a silo and testing it on thursday.

It's probably better than risking any further regressions by doing the apt and packagekit version upgrades.

Changed in packagekit (Ubuntu RTM):
status: New → Fix Released
assignee: Pat McGowan (pat-mcgowan) → nobody
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.