juju-core 1.18.4 MRE tracker

[SRU Notes]

This SRU includes three changes:

1) Upstream version bump from 1.18.1 to 1.18.4: provisional MRE approved by the Technical Board: https://wiki.ubuntu.com/StableReleaseUpdates/MicroReleaseExceptions

2) Packaging change for bug 1325025 (distro-info dependency) has SRU information in that bug.

3) New and enhanced dep8 tests copied from Utopic packaging.

No other changes. This happens to make the source package identical to juju-core in Utopic, differing only in debian/changelog.

Pre-upload testing: dep8 tests against juju-core and juju-quickstart (SRU to be uploaded currently) passed.

SRU verification will include further testing, including juju-quickstart testing as stipulated in the MRE.

juju-quickstart is also being unbroken by both this SRU and a separate juju-quickstart (non-MRE) SRU in bug 1306537, bug 1311321 and bug 1309678. Both SRUs need to be in trusty-proposed together to effectively test juju-quickstart. However, as juju-quickstart is completely broken in Trusty currently, and requires both this SRU and the juju-quickstart SRU to land to fix, there is no need for one SRU to block the other.

The Utopic juju-core powerpc FTBFS is not expected to affect Trusty; it is a suspected toolchain issue that only appears to affect Utopic.

[Description]

1.18.4 contains essential fixes that need to land in Trusty.

Changes extracted from bzr for 1.18.1 Trusty (currently in Trusty) to 1.18.4:

------------------------------------------------------------
revno: 2272 [merge]
author: John Arbash Meinel <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Sat 2014-04-12 06:15:19 +0000
message:
  [r=jameinel] testing/filetesting: preserve bzr file-ids

  filetesting was backported from trunk to 1.18 using diff+patch rather than a bzr merge that would preserve its file identifiers. That means that when merging from 1.18 => trunk all those files look like duplicated files. This patch just fixes up the Bazaar indentifiers so that they match, which should allow easier merging between 1.18 and trunk in the future.

  As this is a metadata only change (no code changes), I'm going to self approve this.
------------------------------------------------------------
revno: 2273 [merge]
author: Curtis Hovey <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Sat 2014-04-12 09:54:13 +0000
message:
  [r=jameinel] Increment juju stable to 1.18.2

  Increment juju stable version and win installer to 1.18.2.

  https://codereview.appspot.com/87120043/
------------------------------------------------------------
revno: 2274 [merge]
author: John Arbash Meinel <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Thu 2014-04-24 14:14:20 +0000
message:
  [r=jameinel],[bug=1311676] worker/provisioner: no tools is an error

  We were treating no tools found as an error of the Provisioner, rather
  than an error with the machine we were trying to provision (bug
  #1311676). This now sets the status of the machine to an error state,
  which lets us continue with our lives.

  In the original bug about not having tools, it was shown that in the
  local provider, doing

    juju bootstrap -e local
    juju deploy -e local precise/ubuntu
    juju deploy -e local trusty/ubuntu ubuntu-t

  Would end up with both machines stuck in Pending. Now it says:
  $ juju status -e local
  environment: local
  machines:
    "0":
      agent-state: started
      agent-version: 1.18.2.1
      dns-name: localhost
      instance-id: localhost
      series: trusty
    "1":
      agent-state-info: '(error: no matching tools available)'
      instance-id: pending
      series: precise
    "2":
      agent-state: started
      agent-version: 1.18.2.1
      dns-name: 10.0.3.194
      instance-id: jameinel-local-machine-2
      series: trusty
      hardware: arch=amd64

  And you can see that it succeeded in deploying trusty because it put
  precise into an error state.

  https://codereview.appspot.com/93720044/
------------------------------------------------------------
revno: 2275 [merge]
author: Nate Finch <email address hidden>, John Arbash Meinel <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Thu 2014-04-24 20:35:41 +0000
message:
  [r=natefinch] juju/api.go: actually handle error (bug #1312136)

  (refactored from a change John made)

  If there is a problem reading the .jenv file, it can return an error,
  but we weren't paying attention to it. Instead we would end up getting a
  panic() a few lines down when we go to use a *Config that is actually
  nil.

  addition by Nate:

  Refactored the code to extract the logic for getting the config into a separate function, which cleans up the error handling significantly.
------------------------------------------------------------
revno: 2276 [merge]
author: Curtis Hovey <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Thu 2014-04-24 21:26:22 +0000
message:
  [r=sinzui] Fix loggo import.

  github.com/loggo/loggo moved to github.com/juju/loggo.

  https://codereview.appspot.com/96780043/
------------------------------------------------------------
revno: 2277 [merge]
author: John Arbash Meinel <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Fri 2014-04-25 11:18:42 +0000
message:
  [r=jameinel],[bug=1312537] Rename authorised-keys to authorized-keys, but keep the other as an alias.

  This makes us a bit more consistent with the value in environments.yaml.
  (Addresses bug #1312537)
  https://codereview.appspot.com/93800043/
------------------------------------------------------------
revno: 2278 [merge]
author: Ian Booth <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Tue 2014-04-29 18:04:13 +0000
message:
  [r=wallyworld],[bug=1302205] Cherry pick trunk 2588,2589 to backport fix for bug 1302205 "
  manual provisioned systems stuck in pending on arm64"
------------------------------------------------------------
revno: 2279 [merge]
author: Ian Booth <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Tue 2014-04-29 18:23:34 +0000
message:
  [r=wallyworld],[bug=1304742] Cherry pick trunk r2600 to fix bug #1304742: version reports "armhf" on arm64
------------------------------------------------------------
revno: 2280 [merge]
author: Martin Packman <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Wed 2014-04-30 19:47:36 +0000
message:
  [r=gz],[bug=1302820] lxc and kvm broker fake tools for other series

  Cherrypick of r2583 from trunk.

  Currently the lxc and kvm brokers only create the same
  series containers as the host, even if a different series
  is asked for.

  Since the series is determined by the tools, we tweak
  the possible tools in order to give the right series.

  Original review: https://codereview.appspot.com/85220043/

  https://codereview.appspot.com/93990043/

  R=dimitern
------------------------------------------------------------
revno: 2281 [merge]
tags: juju-1.18.2
author: Andrew Wilkins <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Wed 2014-04-30 20:30:31 +0000
message:
  [r=axwalk] Set machine addresses during bootstrap

  Fixes lp:1314430

  https://codereview.appspot.com/99950043/
------------------------------------------------------------
revno: 2282 [merge]
author: Jesse Meek <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Wed 2014-04-30 21:27:20 +0000
message:
  [r=waigani],[bug=1301538] local provider: check for juju-local

  Cherrypick r2599 from trunk.

  R=axwalk
------------------------------------------------------------
revno: 2283 [merge]
author: Curtis Hovey <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Wed 2014-04-30 22:32:31 +0000
message:
  [r=sinzui] Increment juju stable to 1.18.3

  Update version and installer to 1.18.3

  https://codereview.appspot.com/92930043/
------------------------------------------------------------
revno: 2284 [merge]
author: Ian Booth <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Fri 2014-05-09 01:51:33 +0000
message:
  [r=wallyworld],[bug=1304407] Backport fix for 1304407
------------------------------------------------------------
revno: 2285 [merge]
author: Andrew Wilkins <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Fri 2014-05-09 03:48:39 +0000
message:
  [r=axwalk],[bug=1316869] Init EnvCommandBase.EnvName from envs.yaml

  This change brings 1.18 in line with trunk,
  where EnvCommandBase.EnvName is initialised
  to the default environment if current-environment
  does not exist and neither "-e" nor $JUJU_ENV
  have been specified.

  The primary change is in moving environcommand.go
  to cmd/envcmd, and adding an Init method that
  reads from environments.yaml as a last resort.
  All commands that embed EnvCommandBase are updated
  to call this method, and various tests are updated
  as a result.

  Fixes lp:1316869

  https://codereview.appspot.com/92140043/
------------------------------------------------------------
revno: 2286 [merge]
author: Ian Booth <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Fri 2014-05-09 04:13:41 +0000
message:
  [r=wallyworld],[bug=1306537] Backport fix for bug 1306537
------------------------------------------------------------
revno: 2287 [merge]
author: Andrew Wilkins <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Fri 2014-05-09 04:30:32 +0000
message:
  [r=axwalk],[bug=1315216] Introduce lxc-use-clone, ContainerManagerConfig

  (Backport from trunk to 1.18)

  A new lxc-use-clone config attribute and a new
  ContainerManagerConfig provisioner API are introduced.
  If lxc-use-clone is set to true, then the LXC
  provisioner will be configured to use cloning
  regardless of provider type. This setting is currently
  immutable; additional work will be required to support
  reacting to changes and reconfiguring the provisioner.

  Fixes lp:1315216
------------------------------------------------------------
revno: 2288 [merge]
tags: juju-1.18.3
author: John Arbash Meinel <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Fri 2014-05-09 09:40:55 +0000
message:
  [r=jameinel],[bug=1301353],[bug=1304770] testing: use juju-mongodb when appropriate

  This should address parts of both bug #1301353 and bug #1304770.
  Instead of requiring a 'mongod' to be in PATH, we'll also search for
  /usr/lib/juju/bin/mongod which is from the juju-mongodb package.
  For core tests, we prefer the juju-mongodb version (since jujud itself
  will prefer it), and for store tests we prefer 'mongod' on PATH since
  it will have the v8 engine and allow us to run all the tests.

  This also auto-detects when noTestMongoJS should be set because it is
  using the /usr/lib/juju/bin/mongod executable. Note that this
  autodetection *fails* for people who have been hacking around support by
  symlinking /usr/lib/juju/bin/mongod back into their PATH. So I didn't
  remove the ENV variable, but it should no longer be necessary.

  I'm targetting 1.18 because with this patch I think Juju CI will have to
  do less hackery to have the test suite JustWork for them.

  https://codereview.appspot.com/97240043/
------------------------------------------------------------
revno: 2289 [merge]
author: Curtis Hovey <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Fri 2014-05-09 16:49:18 +0000
message:
  [r=sinzui] Increment juju stable to 1.18.4

  Update juju version and win installer to 1.18.4.
------------------------------------------------------------
revno: 2290 [merge]
author: Ian Booth <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Mon 2014-05-12 03:38:22 +0000
message:
  [r=wallyworld],[bug=1314686] Backport utopic support from trunk, for bug lp:1314686
------------------------------------------------------------
revno: 2291 [merge]
author: Ian Booth <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Mon 2014-05-12 03:54:56 +0000
message:
  [r=wallyworld],[bug=1312176] Simple fix to update the fake tools upload so that we always include tools for precise and trusty no matter what series the tests are run on.
------------------------------------------------------------
revno: 2292 [merge]
author: Ian Booth <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Tue 2014-05-13 03:39:49 +0000
message:
  [r=wallyworld],[bug=1318729] Another attempt to fix tools tests

  A first attempt to ensure the required fake
  tools were uploaded for tests failed.
  This rework uses a cleaner approach - the
  ToolsLtsSeries slice from environs/bootstrap
  is used to determine what tools to upload.

  https://codereview.appspot.com/97160044/
------------------------------------------------------------
revno: 2293 [merge]
author: Andrew Wilkins <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Tue 2014-05-13 04:08:56 +0000
message:
  [r=axwalk],[bug=1317197] Take hook exec lock in container host init

  When initialising the host for a container type,
  we will install packages (lxc, kvm, etc.); this can
  conflict with hook execution. This CL modifies the
  container initialisation code to acquire the hook
  execution lock when initialisting the host to avoid
  the conflict.

  Fixes lp:1317197
------------------------------------------------------------
revno: 2294 [merge]
author: Andrew Wilkins <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Thu 2014-05-22 02:21:23 +0000
message:
  [r=axwalk],[bug=1321025] Use the right mongod binary on utopic

  There was some broken code in agent/mongo
  for calculating the mongod path. It special
  cased trusty in two different ways.

  Fixes lp:1321025

  https://codereview.appspot.com/99410053/
------------------------------------------------------------
revno: 2295 [merge]
author: Ian Booth <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Mon 2014-06-02 08:00:19 +0000
message:
  [r=wallyworld],[bug=1325072] Fix unit tests on utopic

  Backport work to fix unit tests on utopic.

  https://codereview.appspot.com/105750043/
------------------------------------------------------------
revno: 2296 [merge]
tags: juju-1.18.4
author: Ian Booth <email address hidden>
committer: Tarmac
branch nick: 1.18
timestamp: Mon 2014-06-02 23:35:41 +0000
message:
  [r=wallyworld] Fix root-disk consraints on ec2

  EC2 instance types do not include a root
  disk size as this is set on instance
  creation. This was causing the contraints
  matching to fail when root disk was specified.
  The fix is to ignore 0 root disk values when
  matching constraints.

  https://codereview.appspot.com/106750044/
------------------------------------------------------------
Use --include-merged or -n0 to see merged revisions.