add-machine lxd:0 provisioning error: "Error adding alias ubuntu-xenial: not found"

[Andrew McDermott]

Tried to add a container with:

  $ juju add-machine lxd:0

currently errors with:

$ juju show-machines
model: maas
machines:
  "0":
    juju-status:
      current: pending
      since: 16 Mar 2016 18:03:17Z
      version: 2.0-beta3.1
    dns-name: 10.17.20.215
    instance-id: /MAAS/api/1.0/nodes/node-562d8f0c-d8e4-11e5-b0d8-525400c38f67/
    machine-status:
      current: running
      message: Deployed
      since: 16 Mar 2016 18:03:30Z
    series: xenial
    containers:
      0/lxd/0:
        juju-status:
          current: error
          message: |
            Error adding alias ubuntu-xenial: not found
          since: 16 Mar 2016 18:19:05Z
        instance-id: pending
        machine-status:
          current: provisioning error
          message: |
            Creating container: Error adding alias ubuntu-xenial: not found
          since: 16 Mar 2016 18:19:04Z
        series: xenial
    hardware: arch=amd64 cpu-cores=1 mem=1024M
    controller-member-status: has-vote

A workaround is to run the following on the machine node:

  $ lxd-images import ubuntu --alias ubuntu-xenial xenial

which results in an alias name in the following output:

ubuntu@maas19-node2:~$ lxc image list
+---------------+--------------+--------+---------------------------------------------+--------+----------+------------------------------+
| ALIAS | FINGERPRINT | PUBLIC | DESCRIPTION | ARCH | SIZE | UPLOAD DATE |
+---------------+--------------+--------+---------------------------------------------+--------+----------+------------------------------+
| ubuntu-xenial | 618ab2ddf554 | no | ubuntu 16.04 LTS amd64 (beta1) (20160223.1) | x86_64 | 139.53MB | Mar 16, 2016 at 6:43pm (UTC) |
+---------------+--------------+--------+---------------------------------------------+--------+----------+------------------------------+

[John A Meinel (jameinel)]

Can you tell us the git version of your Juju? We've landed several patches recently that automatically do the import of images from cloud-images.ubuntu.com

[Andrew McDermott (frobware)]

 07693b816c207bc6e3fa9d7dd3f76784a695908e on maas-spaces-multi-nic-containers. This is a little behind master.

[Andrew McDermott (frobware)]

This is no longer an issue for me on the maas-spaces-multi-nic-containers-with-master branch at commit:

  5ade2a6117aae4be18356548f4e25453df2f1afe

[Launchpad Janitor (janitor)]

[Expired for juju-core because there has been no activity for 60 days.]

[John A Meinel (jameinel)]

I just ran into this with juju 2.0.3 on Xenial.

My first guess is that I *do* have an image with alias ubuntu-xenial already, however, because we prefer the "upstream" definition of container images, we are downloading a replacement, but then telling LXD to use the alias which collides with the existing alias.

This is with LXD 2.8 from xenial backports.

Its possible that https://github.com/juju/juju/pull/6970 fixes this, as I'm currently explicitly testing 2.0.3.

[John A Meinel (jameinel)]

It looks like one possible trigger for this, is that I was using "release" as my 'ubuntu-xenial' but conjure-up triggers us to use "daily" which would be a different source, and thus more likely to trigger this, as otherwise we're likely to already have the same thing as upstream as we do "auto-update".