When no root_device hint is set, an MDRAID partition can be incorrectly
selected as the root device which causes installation of the bootloader
to the physical disks behind the MDRAID volume to fail. See the notes
in the referenced Story for more detail.
This change adds a little more specificity to the listing of block
devices.
Change-Id: I66db457e71a0586723ee753bef961aec5bf58827
Story: 2007905
Task: 40303
delete_configuration still fetches all devices as it needs to clean
ones with broken RAID.
Story: #2007907
Task: #40307
Change-Id: I4b0be2b0755108490f9cd3c4f3b71a5e036761a1
Caches hardware information collected during inspection
so that the initial lookup can occur without any delay.
Also adds logging to track how long inventory collection takes.
Co-Authored-By: Dmitry Tantsur <dtantsur@protonmail.com>
Change-Id: I3e0d237d37219e783d81913fa6cc490492b3f96a
For some reason pep8 test started to complain causing mayhem.
This patch fixes the issues and does some refactor of dmi_inspector
tests moving pure data to a separate file.
Change-Id: Ia244a496acd80abad679f8ae9832d4f0471500e7
This function checks for /sys/firmware/efi. Some tests do not mock
isdir, so they fail on UEFI machines.
Change-Id: I088218ddb88717ac07669d0b97c6cd50208ede8c
The proposed changes concern two steps:
First, when creating the RAID configuration, have a GPT partition
table type (this is not necessary, but more natural with UEFI).
Also, leave some space, either for the EFI partitions or the BIOS
boot partitions, outside the Software RAID.
Secondly, when installing the bootloader, make sure the correct
boot partitions are created or relocated.
Change-Id: Icf0a76b0de89e7a8494363ec91b2f1afda4faa3b
Story: #2006379
Task: #37635
Now that an operator can pick the devices that participate in RAID,
it no longer makes sense to verify all devices.
Change-Id: Id5d8d539183f0db4ba3c4132ce6bc9919f9cd1ea
Story: #2006369
This change adds support for the physical_disks RAID parameter in
a form of device hints (same as for root device selection).
Change-Id: I9751ab0f86ada41e3b668670dc112d58093b8099
Story: #2006369
Task: #39080
In the interest of keeping CI happy, and reducing the need for
wasteful rechecks, I've changed a test which was increasingly
seeing unit test failures on py37 and py38 testing.
This test functionally still does what is required, validates only
one thread would be utilized by default, but doesn't do output
validation since the calls, threading, and mock interaction were
leading to test failures.
Change-Id: Ia658e0aba001d6195cafeb4484b176e9147e3c26
Pyudev is used to return extra data about a device
using the udev interface.
Sometimes that lookup doesn't quite work, like on
md devices after restarting them. As such, we will
now tollerate the failure and continue the process
as before.
Change-Id: Ibbc1759fe2cd3d7d09019b4e80d3c61d54c844dd
Story: 2007281
Task: 38726
HPE "Virtual Install Devices" appear as read-only block
devices, and may... or may not be visible depending on the
bios configuration state.
These devices can no longer be disabled from the bios settings
so the simplest course of action seems to be that we should
handle the existence of a read-only device.
In the event of secure erase, this is treated as a hard failure
case and a driver_internal_info flag has been added to enable
a future bypass method for knowledgable operators.
Change-Id: Ief8b360d11e654d8fae3a04a2a9f8d474a06e167
Story: 2007229
Task: 38502
This change adds a new introspection data field 'configuration'
with two lists: managers and collectors.
Change-Id: Ice0d7e6ecff3f319bc3a4f41617059fd6914e31c
This patch fixes the call to ApplyResult from the multiprocessing
pool module.
The error presents itself more clearly when running tests using
Python 3.8, and that's why it's not been noticed so far.
The change in behavior in Python 3.8 will raise the error:
AttributeError: 'dict' object has no attribute '_cache'
Change-Id: I6717ebf5fef0d32120c9386ab8e47cc81f328493
For the future deployment API we need to be able to set root_device
per deployment in addition to per node. This change adds it.
Change-Id: I1dd046c2e5fca211a84290bac8daa7550b21614f
Story: #2006910
Task: #37955
Since we've dropped support for Python 2.7, it's time to look at
the bright future that Python 3.x will bring and stop forcing
compatibility with older versions.
This patch removes the six library from requirements, not
looking back.
Change-Id: I4795417aa649be75ba7162a8cf30eacbb88c7b5e
Looks like adding RAID 5/6 support may be easier than most
could immagine. The code, as written appears to be safe and
logical creating a RAID 5 or RAID 6 volume.
Not that we can really test this in CI, but it seems only
validation code needs to be changed to loosen the constraint.
Change-Id: Ib891b3c97f0bfb02af3b59581a451c4b25e03b85
This reverts commit 258d963e406c512bb90295c700ee22e4609abcd0.
Remove the mbr/gpt choice from softraid features for now, as it cannot be
directly used without additional commits, to be pushed soon. Furthermore, it
could even lead to instance spawn misconfiguration, if the disk label
specified in instance_info cannot fit with the boot mode + partitioning layout
(example: you build softraid over gpt, then you lose the mbr gap. Thus you need
an additional bios boot partition, in BIOS boot mode, or an esp, in UEFI boot
mode, outside the raid, for grub to install itself inthere, to be able to
assemble raid and find root device correctly).
Change-Id: I3a0a704ea99a40eb3fc8e879270dfbd356951488
Cleaning presently fails on floopy disk controllers.
While they may be uncommon in server hardware, they
can exist in virtual machines, and even as virtual
devices on some hardware chassises.
Change-Id: I8ba07bfd5ca1e503f46c1bac4fffb5f509186939
Story: 2006419
Task: 36309
With the addition of devices with type 'MD' during
list_all_block_devices (for "empty" devices), partitions on top
of md devices are now returned as well. These don't need deletion,
so skip these during software RAID cleanup.
Story: #2006355
Task: #36121
Change-Id: Idff9b9b0cbe42b57aa315ae07c0b51c35c9ade1e
Fixes:
- Softraid creation: size_gb unit conversion to default parted unit.
- Softraid creation issue when we specify several logical disks with
more than one having its size specified (not 'MAX').
- Softraid creation issue when a logical disk with size_gb 'MAX' is
not last in the list.
Change-Id: I13e087611354a35acc2925037f1b10047477d972
Story: 2006352
Task: 36118
While most modern Linux distributions do not create RAM disks, newer
TinyCoreLinux actually does. Ignore them since we don't want to deploy
to RAM.
Change-Id: Ia810c7fb47f7098b8173d765ff9c33b48a223ff1
These devices are mapped into memory, but nonetheless can appear
in lsblk output. Ignore them explicitly.
Change-Id: I2bb595ef29abdc98bd887bdf41c3c7f64cc3a7e6
The output for aarch64[1] `lshw -quiet -json` is a bit different from
x86_64, this will lead to 0 physical memory calculated currently. This
patch aims to make it compatible for aarch64 when getting physical memory
[1]: http://paste.openstack.org/show/754789/
Change-Id: Ib7b4ee857d8550e3d75ac8a53c61f0ecab45feff
