lp:~xnox/curtin/full-uuid

Branch merges

Propose for merging

No branches dependent on this one.

Merged into lp:~curtin-dev/curtin/trunk

Server Team CI bot: Approve (continuous-integration) on 2017-03-07

Scott Moser (community): Needs Fixing on 2017-02-06

Diff: 25 lines (+9/-3)

1 file modified

curtin/commands/block_meta.py (+9/-3)

api

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Related blueprints

360. By Dimitri John Ledkov on 2016-03-04

Use UUID, if available, for full-disk*, partition or dm-crypt volumes.

Previously, if a full-disk was formatted with a filesystem
(e.g. btrfs) instead of mounting by UUID, it was mounted by path
(/dev/vdd). However, a uuid can be used instead.

359. By Scott Moser on 2016-02-25

tools/ssh-keys-list: fix a bug in modes, add argparse

There was actually no way to request 'json' output. And the mode
'public-keys' would result in a stack trace.

This fixes those 2 issues and adds argparse so we get some help.

358. By Ryan Harper on 2016-02-24

Add bond parsing & improved source, source-directory parsing for /e/n/i.

This give the ability for the parser to handle bond interface in the
/etc/network/interfaces configuration. The resulting dictionary will contain a
bond dictionary if any bond options are defined for that interface definition
in /etc/network/interfaces.
Improves the source and source-directory parsing to match was is documented in
/etc/network/interfaces for ubuntu.
Supports parsing glob in both source and source-directory and performs regex
matching for source-directory to parse the same as /etc/network/interfaces is
defined. A _source_path key is provided in each dictionary entry for each
element defining the path the interface was created.

357. By Scott Moser on 2016-02-24

tox.ini: run with 'python3' rather than python3.4 or python3.5

There is python3 on xenial, but it is python3.5 which differs
from the python3.4 which is on trusty.

This change still allows 'tox' to work on trusty.
Admittedly, we do test python3.5 only on xenial and python3.4 only
on trusty, but we can solve that at a future date.

356. By Ryan Harper on 2016-02-23

curtin/net: move global dns-* options under auto lo in ENI

ifupdown expects any dns-* options to be related to an iface stanza
The only iface ensured to be called with ifup is lo; moving the dns-*
parameters under iface lo allows dns options to always be present even
if other interfaces are not ifup'ed.

Update unittests and add new vmtest (TestNetworkStatic) to confirm
proper configuration under lo iface.

355. By Ryan Harper on 2016-02-23

vmtests: implement uefi boot mode via pflash

Add support for uefi to tools/launch and teach vmtests to switch to uefi
pflash mode if requested. vmtests generates and supplies the writable
nvram file that's passed to launch now. The basic uefi test currently
checks that expected files are in /sys/firmware/efi/*. Handle
older combined OVMF format found on trusty hosts as well as the newer
split firmware in Wily and newer.

354. By Scott Moser on 2016-02-23

vmtest: fix regression in psuedo tests

The psuedo test class was broken by the last commit.
This is a simple fix for:
  nosetests3 tests/vmtests/test_vmtests.py

353. By Scott Moser on 2016-02-22

Hardware Enablement Stack (HWE) Tests for Trusty and Precise

This series strives for better vmtest coverage of HWE Kernels by:
- Provide infrastructure for HWE Tests
- Defining Trusty-HWE-[uvw] in various tests (all working).
  - Basic
  - Bonding
  - Network
  - Raid5Bcache
  - LVM
  - We didn't define all (where we considered them covered otherwise)
  - we did not enable all of them as a tradeoff to CI test execution time.
- Define minimal Precise-HWE-T Tests
  - enabled one Disk case (Bcache basic)
  - enabled one Net case (bonding)
  - some more defined but disabled (comment lists last known issue)

352. By Scott Moser on 2016-02-17

vmtest: when putting console to debug log, decode with errors=replace

The console logs from boot and install are not guaranteed to have
utf-8 decodable content written to them. This would cause a stacktrace
on the failure to decode.

In practice, I saw this when testing precise while working on UEFI.

351. By Scott Moser on 2016-02-12

partitioning: limited support for odd ordering of partition numbers

This fixes a bug in the partitioning code where it assumes that the
previous partition is numbered N-1. That is not always the case.
The change here is very focused on:
 a.) Being non-invasive.
 b.) fixing bug 1543263 and allowing maas to send configuration
     for power that "hides" the PReP partition as number 15
     so they don't have to really show it to the user.

Ultimately, we want to re-write the partitioning code to walk
through the storage config and correctly fill in number, start
and end for each entry. That is a more invasive and time-consuming
change though.

This also adds another disk to examples/tests/basic.yaml
and tests that it is partitioned as expected in
tests/vmtests/test_lvm.py