UTAH fails to install oem with raring desktop

Bug #1087679 reported by Para Siva
12
This bug affects 2 people
Affects Status Importance Assigned to Milestone
UTAH
Fix Released
Critical
Unassigned

Bug Description

The raring desktop OEM preseeded installations are failing and it could be noticed that the installation does not proceed after the reboot step in oem installations.

Attached is the syslog from the failed job.

The impacted jobs are,
https://jenkins.qa.ubuntu.com/view/Raring/view/Smoke%20Testing/job/raring-desktop-amd64-smoke-oem/7/console
https://jenkins.qa.ubuntu.com/view/Raring/view/Smoke%20Testing/job/raring-desktop-i386-smoke-oem/4/console

Though these jobs have the post installation test results, the installation itself is not complete and the post installation tests seem to have run before the full oem installation

Manual OEM installations on hardware and VMs are working fine and even preseeded installations without using utah are working fine.
Please see attached file called 'passed' for the syslog of a passed preseeded installation

Related branches

Revision history for this message
Para Siva (psivaa) wrote :

The syslog from a failed image

tags: added: iso-testing qa-daily-testing rls-r-incoming
description: updated
Revision history for this message
Para Siva (psivaa) wrote :

Passed syslog for preseeded oem installation without using UTAH for comparison if needed

description: updated
Para Siva (psivaa)
Changed in utah:
importance: Undecided → Critical
Gema Gomez (gema)
tags: removed: rls-r-incoming
Revision history for this message
Para Siva (psivaa) wrote :
Revision history for this message
Max Brustkern (nuclearbob) wrote :

Adding these two lines to the preseed made one test pass:
d-i oem-config/enable boolean true
d-i oem-config/id string USIT_OEM
The remaining failure was this:
  stderr: |-
    INFO: Checking if desktop shortcut (/home/oem/Desktop/oem-config-prepare-gtk
.desktop) exists
    F
    ======================================================================
    FAIL: test_desktop_shortcut_exists (__main__.DesktopShortcutExistTest)
    ----------------------------------------------------------------------
    Traceback (most recent call last):
      File "./test.py", line 37, in test_desktop_shortcut_exists
        "Desktop shortcut doesn't exist")
    AssertionError: Desktop shortcut doesn't exist

    ----------------------------------------------------------------------
    Ran 1 test in 0.001s

    FAILED (failures=1)
I don't see that symlink somewhere else, so I'm going to see if using the oem user or at least creating that home directory will make a difference.

Revision history for this message
Max Brustkern (nuclearbob) wrote :

I've been testing with an alternate config file that changes the user to 'oem' instead of 'utah'. The first thing I found with this is that the sshprivatekey defaults to the oem user's home directory, so I had to change that to point to the utah user in the same config file. After that, I realized the uploadfiles method of the SSHMixin class doesn't specify a username, so all files are uploaded as the user running the script, in this case, utah. This fails when there is no utah user, but confusingly throws this exception:
paramiko.SSHException: not a valid DSA private key file
I'm going to test a utah code change to see if that in conjunction with the updated config file and the preseed changes will fix this.

Revision history for this message
Max Brustkern (nuclearbob) wrote :

I've submitted an additional merge proposal to allow the use of launchpad URLs for utah config files. I'm testing these changes now. If they work for this test, I'll propose a merge against the desktop tests branch that updates the preseed and adds the appropriate config file, and updates the job generation script to include this file.

Changed in utah:
status: New → Triaged
Changed in utah:
status: Triaged → In Progress
Changed in utah:
status: In Progress → Fix Committed
status: Fix Committed → Fix Released
status: Fix Released → In Progress
Revision history for this message
Javier Collado (javier.collado) wrote :

Just realized that the build is broken because the import for url_argument
fails when using `config.py` at build time.

Changed in utah:
status: In Progress → Fix Committed
Revision history for this message
Paul Larson (pwlars) wrote :

Quick update, as of today, the only remaining failure related to this is the missing desktop link for preparing shipping to end user. I manually checked this on todays image and confirmed that the desktop link *is* present as it should be. Talked to Max this morning and he said it's just because we're using the utah user instead of the oem user it expects. He has a fix coming for that soon.

Changed in utah:
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.