[SRU] livecd-rootfs ubuntu-cpc vagrant image builder

Bug #1605795 reported by Louis Zuckerman
146
This bug affects 23 people
Affects Status Importance Assigned to Milestone
cloud-images
New
Undecided
Unassigned
livecd-rootfs (Ubuntu)
Confirmed
Undecided
Brian Murray

Bug Description

[Impact]

* The official ubuntu vagrant box for xenial has the following problems:
- synced folders don't work (LP: #1565985)
- the default synced folder is disabled
- only one instance of the box can be run on a host
- localhost name resolution fails (LP: #1561250)

* These issues are all fixed in yakkety by my patch [MR]. We would like to backport the changes to xenial so that the vagrant boxes for the LTS release will work correctly.

* The way this fix works is by slightly changing how the vagrant box is built. Instead of using the base VMDK image and making a box from that, we use the ext4 image, install the virtualbox-guest-utils package, then create our own vmdk for the vagrant box. The other changes were made to config files included as vagrant box metadata alongside the vmdk image.

[Test Case]

* Test instructions are explained in comments [MR.1] and [MR.2] in the original merge request [MR]

* The change was tested and verified on xenial by myself (the original patch author).

* The change was also tested on yakkety by Dan Watkins using the actual ubuntu-cpc image building platform and verified by users who downloaded the box Dan built. (See LP: #1565985 for their verifications.)

[Regression Potential]

* I believe there is little chance of regression because this patch...
- Only affects livecd-rootfs which is a package that is used to build ubuntu distribution disk images, and not needed for running an ubuntu installation
- Only changes one image builder script within livecd-rootfs, ubuntu-cpc/hooks/042-vagrant.binary, which no other builders depend on

* The only possible regression I'm aware of is that the build could fail if there is not enough free space. This patch causes the build to use about 2 GB more space than previously because an extra disk image is created during the build process (then cleaned up after).

[Other info]

* This patch was developed in collaboration with Dan Watkins (with review & feedback also by Steve Langasek and Adam Conrad), and I'm grateful for their help.

* Dan advised me in [IRC] to create this SRU and said that someone will need to target LP: #1565985.

[Links]

[MR] https://code.launchpad.net/~semiosis/livecd-rootfs/fix-for-1565985/+merge/298305
[MR.1] https://code.launchpad.net/~semiosis/livecd-rootfs/fix-for-1565985/+merge/298305/comments/768010
[MR.2] https://code.launchpad.net/~semiosis/livecd-rootfs/fix-for-1565985/+merge/298305/comments/770577
[IRC] https://irclogs.ubuntu.com/2016/07/22/%23ubuntu-devel.html#t15:24

no longer affects: cloud-archive
description: updated
description: updated
Revision history for this message
Louis Zuckerman (semiosis) wrote :
Revision history for this message
Launchpad Janitor (janitor) wrote :

Status changed to 'Confirmed' because the bug affects multiple users.

Changed in livecd-rootfs (Ubuntu):
status: New → Confirmed
Revision history for this message
Chris Pick (cpick) wrote :

What can I do to help move this ticket forward?

The fix was identified and applied to newer releases months ago.

Lack of ease of local, Vagrant-based testing has impeded my deployment of Xenial to my production systems.

Revision history for this message
Daniel Lynch (dlynch15-8+lp) wrote :

I just ran into this issue as well. Is there anyone providing a working xenial64 image?

Revision history for this message
Daniel Lynch (dlynch15-8+lp) wrote :

I saw a suggestion (sorry, forget who), to use bento/ubuntu-16.04 ( https://atlas.hashicorp.com/bento/boxes/ubuntu-16.04 ). It's made by the people who make Chef.

This seems to work fine, I guess it's what I'll be using.

Steve Langasek (vorlon)
Changed in livecd-rootfs (Ubuntu):
assignee: nobody → Brian Murray (brian-murray)
Revision history for this message
Brian Murray (brian-murray) wrote :

Louis - I'm going to start working on backporting your changes to 16.04. The changelog will end up referencing the two bugs in the description and not this bug, so it'd be good if test cases (testing the specific change e.g. hostname resolution) existed in each bug's individual description. Thanks!

Revision history for this message
Louis Zuckerman (semiosis) wrote :

Added brief test instructions to LP #1565985. The hostname bug, LP #1561250, already has pretty concise description with how to test.

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.