~raharper/curtin:fix/tox-py27-on-focal

Last commit made on 2020-03-23
Get this branch:
git clone -b fix/tox-py27-on-focal https://git.launchpad.net/~raharper/curtin
Only Ryan Harper can upload to this branch. If you are Ryan Harper please log in for upload directions.

Branch merges

Branch information

Name:
fix/tox-py27-on-focal
Repository:
lp:~raharper/curtin

Recent commits

e1b7166... by Ryan Harper on 2020-03-23

tox.ini: Fix issues with newer tox on focal

Newer tox on focal is not happy when inhieriting dependencies.
First in py27 testenv, it does not pickup the 'testenv' deps
which includes test-depedndences and dependencies files. Fix
this by explicitly referencing these deps fir testenv:py27.

Second, in testenv:trusty-p27, which recreates specific package
versions to model the trusty build environment we cannot pull in
the default testenv deps, as this breaks with different versions
of mock and other settings. Instead, just bring in the setuptools
dep to ensure we can keep running on py27 for now.

b10aeef... by Ryan Harper on 2020-03-12

vmtest: Fix test_basic.py to run on s390x

The TestBasic currently fails on s390x due to usage of virtio-disks
and an "empty" virtio-scsi device controller. There is an existing
bug in s390x firmware that does not like an "empty" virtio-scsi
controller. Until that bug is fixed, skip TestBasic on s390x and
instead make TestScsiBasic work. This requires capturing output of
/dev/disk/by-path as s390x fstab entries prefer by-path as paths on
s390x are stable[1].

1. https://wiki.ubuntu.com/FSTAB

LP: #1866663

90bc0d7... by Ryan Harper on 2020-03-11

vmtest: use util.load_file for loading collect files

In some cases the content inside collect files may need to be
decoded. Instead of open-coded read and decode, use the common
util.load_file() which already handles decoding data.

- Drop mode= param, no callers used it
- Reuse load_collect_file() for load_collect_file_shell_content
- load_log_file callers specify full path to file

bf03e4f... by Ryan Harper on 2020-03-10

block-meta: refactor storage_config preserve and wipe settings

The subiquity installer has use-cases where users wish to preserve
some devices and then repurpose them into other storage devices.
For example, retain a particular partition and then add it to an
LVM setup. Curtin currently makes users chose to preserve devices
wholesale or chose to wipe them. To support the 'repurpose' use
case we need to be able to preserve a device and then optionally
wipe the data therein independently of preservation.

The block schema has been updated to accept preserve and wipe
settings for each block device type. For the following types,
dasd, disk, partition, format, lvm_volgroup, lvm_partition,
dm_crypt, raid and bcache, Curtin will accept the preserve
boolean, and if set True, will verify various characteristics
of the target device and compare this to the specified config.

For example, preserving a partition which needs a device, size
and some optional values, like flag. Curtin will verify the
device exists, the number is correct, the size of the partition
matches and if the correct flag is set on the device.

The set of verifications performed are storage type specific.
Any verification failure results in Curtin raising a RuntimeError
exception which includes the expected and found values.

Additional work:

- Migrate bcache creation functions from block-meta into
  block.bcache
- Unittest and vmtests added for verifying preserve for each
  storage type.

LP: #1837214

1f9c53a... by Ryan Harper on 2020-03-09

block-discover: skip 'multipath' key in blockdevice if mpath name is None

Avoid emitting invalid storage-config yaml by not including a 'multipath'
key in block device output without a valid (not None) value for the key.

c650f6c... by Ryan Harper on 2020-03-05

tox: all py27 environments should use the base py27 deps

We fixed base py27 deps to pin setuptools < 45 to fix an issue
with newer setuptools and py27 deprecation. We need source the
base py27 env deps in all py27 related environments

eacb655... by Ryan Harper on 2020-03-04

uefi: refactor efibootmg handling to support removing duplicate entries

When curtin is reusing an existing ESP, it may already contain a previous
ubuntu entry. efibootmgr complains about duplicate labels, but does not
provide a mechanism to remove the duplicate automatically. In curtin
we use our efibootmgr output parser to find duplicates and remove them.

To enable this we need to ensure that 'efivarfs' is mounted inside the
target. In ChrootableTarget we will add the additional mount automatically
if we're running on a uefi system.

This branch also refactors use of ChrootableTarget context manager to
minimize the number of chroot calls (which include 4 mounts and unmounts).

LP: #1864257

d3cac6f... by Ryan Harper on 2020-03-03

tox: pin setuptools < 45 to allow installing py27 in virtenv

setuptools release 45 broke everyone who installed py27[1]. Fix
this by using the previous release.

1. https://github.com/pypa/setuptools/issues/1963

437caaa... by Ryan Harper on 2020-02-26

block-discover: add dasd parser and emit dasd storage config

Probert now includes a 'dasd' element in the probe data.
If present, block-discover will now parse this data and
emit type:dasd elements in the rendered storage config.

LP: #1862849

82f23e3... by Ryan Harper on 2020-02-14

install-grub: refactor uefi partition/disk searching

Refactor the code used to determine the partition and disk where the
ESP resides. There were a few issues:

 - Disks and partitions are of type 'block' which requires the use of
   [ -b ] or [ -e ]; [ -f ] returns false for block devices.
 - The use of lsblk -on pkname exposed a bug in util-linux where pkname
   did not return a value.

This branch adds additional debug logging and safety checks for any
calculated value and fails on empty values. We replace use of lsblk
with use of sysfs to calculate the parent disk of the partition. Add
a vmtest to verify the path where we use efibootmgr to create an entry.

LP: #1862846