| Age | Commit message (Collapse) | Author |
|---|
|
This was painful, but it finishes a TODO from cloudinit/subp.py.
It moves the following from util to subp:
ProcessExecutionError
subp
which
target_path
I moved subp_blob_in_tempfile into cc_chef, which is its only caller.
That saved us from having to deal with it using write_file
and temp_utils from subp (which does not import any cloudinit things now).
It is arguable that 'target_path' could be moved to a 'path_utils' or
something, but in order to use it from subp and also from utils,
we had to get it out of utils.
|
|
Addresses "Runtime Error: dictionary keys changed during iteration".
Co-authored-by: Noah Meyerhans <noahm@debian.org>
LP: #1868327
|
|
|
|
|
|
Fix a typo in a RuntimeError path. (udevamd -> udevadm).
|
|
The cloud-init-local.service expects that any network device name changes
have already been completed by the kernel or udev daemon.
In some situations we've found that the renaming of interfaces from kernel
names (eth0, eth1, etc) to their persistent names (eno1, ens3, enp0s1,
etc) may happen after cloud-init-local has started where it reads values
from sysfs about what network devices are present, and which device to use
as a fallback nic.
Subsequently, cloud-init-local would write out network configuration for a
kernel device name which would no longer be present by the time that
networking services start to bring up the devices. The result is that the
instance does not get networking configured. Prior to use of
systemd-networkd, the Ubuntu 'networking.service' unit included a call to
udevadm settle which is why this race is not seen on a Xenial system.
This change adds the ability to detect if an interface has a stable name,
if if we find one without stable names and stable names have not been
disabled (net.ifnames=0 in /proc/cmdline), then cloud-init will invoke
udevadm settle.
LP: #1766287
|
|
This branch resolves lints seen by pylint revision 1.8.1 and updates our
pinned tox pylint dependency used by our tox pylint target.
|
|
This attempts to use udevadm settle to wait until devices have been
fully "realized". If a device exists, there may still be events in
the udev queue that would create its partition table entries.
We need to wait until those have been processed also.
LP: #1692093
|
|
This fixes several shortcomings of disk_setup with gpt disks.
* 'sgdisk -p' was being used to determine the size of a disk.
this can fail if it believes there is a bad gpt partition table.
Instead we just use blockdev now for both mbr or gpt disks.
* parsing of sgdisk -p output assumed that the 'name' of the partition
type would not have any spaces (Microsoft basic data)
* interaction with sgdisk did not realize that sgdisk wants input
of '8300' rather than '83' and will output the same.
LP: #1692087
|
|
Large instance types have a different disk format on the newly
partitioned ephemeral drive. So we have to adjust the logic in the
Azure datasource to recognize that a disk with 2 partitions and
an empty ntfs filesystem on the second one is acceptable.
This also adjusts the datasources's builtin fs_setup config to remove
the 'replace_fs' entry. This entry was previously ignored, and confusing.
I've clarified the doc on that also.
LP: #1686514
|
|
If 'cmd' is provided to a fs_setup entry, then cloud-init was trying
to execute the rendered string as a single name, rather than
splitting the string. The change here will pass the string to
shell for interpretation so that it is split there.
Also fix some documentation errors and warn when fs_opts or overwrite
is provided along with 'cmd'.
LP: #1687712
|
|
This will change all instances of LOG.warn to LOG.warning as warn
is now a deprecated method. It will also make sure any logging
uses lazy logging by passing string format arguments as function
parameters.
|
|
Accordingly to the documentation:
The ``partition`` option may also be set to ``auto``, in which this
module will search for the existance of a filesystem matching the
``label``, ``type`` and ``device`` of the ``fs_setup`` entry and
will skip creating the filesystem if one is found.
However, using this "auto" flag always recreates the partition no matter
if it has been done before or not.
This commit fixes a bug in which the "partition" attribute was always
set to None although in some cases it should not.
LP: #1634678
|
|
This has been a recurring ask and we had initially just made the change to
the cloud-init 2.0 codebase. As the current thinking is we'll just
continue to enhance the current codebase, its desirable to relicense to
match what we'd intended as part of the 2.0 plan here.
- put a brief description of license in LICENSE file
- put full license versions in LICENSE-GPLv3 and LICENSE-Apache2.0
- simplify the per-file header to reference LICENSE
- tox: ignore H102 (Apache License Header check)
Add license header to files that ship.
Reformat headers, make sure everything has vi: at end of file.
Non-shipping files do not need the copyright header,
but at the moment tests/ have it.
|
|
The version of sfdisk in wily (and onwards) only accepts sectors as a
valid disk size. As such, this refactors the MBR code path in
cc_disk_setup to use sectors.
- use --unit=S: while newer versions of sfdisk assume --unit=S, older
versions do not so we specifically pass it in. Versions of sfdisk
found in supported OSes such as centos6 wont assume --unit=S.
- add --force: this exists back to centos 6 (2.17.2), so it should
be fine, and is what we ultimately want.
"do what I say, even if it is stupid"
- keep --Linux. Even though this has been deprecated for quite some
time, we keep it until versions that want it are unsupported.
If necessary at some point we could check for util linux version
and if it had --Linux and use it in those cases.
Additionally, improve usefulness of some log messages.
LP: #1460715
|
|
The function exec_mkpart_gpt was simply not waiting for udev events
to flush after calling sgdisk. The corresponding function
exec_mkpart_mbr already did.
This should fix a transient failure where mkfs would fail with
'not a block device'.
LP: #1626243
|
|
This adds lots of config module documentation in a standard format.
It will greatly improve the content at readthedocs.
Additionally:
* Add a 'doc' env to tox.ini
* Changed default highlight language for sphinx conf from python to yaml
most examples in documentation are yaml configs
* Updated datasource examples to highlight sh code properly
|
|
If device has no partition table, the first line of output from `sgdisk
-p <device>` will be "Creating new GPT entries.", instead of something
like "Disk /dev/sdb: 266338304 sectors, 127.0 GiB".
Also, protect against localized output by adjusting subp calls that
parse sgdisk output to set LANG=C.
|
|
|
|
Send the --force flag to mkfs or other filesystems when target
is a block device. This fixes a general code flow issue where
we were setting the --force flag.
LP: #1548772
|
|
|
|
|
|
|
|
Fixes Launchpad bug #1311463.
|
|
|
|
MBR uses block sizes, which is what the current (apparently portable) code was
producing. GPT uses sectors to determine partition size.
|
|
|
|
This includes moving some shared logic in to check_partition_layout.
|
|
|
|
This just removes comments '# pylint:' things and other code
remnents of pylint.
|
|
Fixed all complaints from running "make pep8". Also version locked
pep8 in test-requirements.txt to ensure that pep8 requirements don't
change without an explicit commit.
|
|
|
|
|
|
formating support.
|
|
|
|
|
|
makes it cloud agnostic.
|
|
|
|
I'm pretty sure the previous code wasn't seeking correctly
and probably writing near the end, but not to the end.
This is simpler and probably faster.
|
|
|
|
Modified cc_mounts to identify whether ephermalX is partitioned.
Changed datasources for Azure and SmartOS to use 'ephemeralX.Y' format.
Added disk remove functionally
|
|
|
|
|
|
Since for a string there is no difference, we're just
checking for this here.
|
|
|
|
|
|
|
|
this adds 2 functions
update_disk_setup_devices
update_fs_setup_devices
Which update the appropriate datatype, and translate the names.
Translating early means we don't have to deal with updating in the mkfs or
mkpart calls explicitly.
These are more easily unit tested as they just take a dictionary of the
expected type and a 'transformer' that should return a new name or None.
|
|
|
|
|
