| Age | Commit message (Collapse) | Author | Files | Lines |
|
[tool.setuptools.data-files] does not support recursive directory
installation. The old setup.cfg [files] worked because PBR's
FilesConfig.expand_globs() used os.walk() to expand docker/*, but
this hook is not triggered by the pbr.build PEP-517 backend.
Move docker/ into kolla/ and use [tool.setuptools.package-data]
instead. Update kolla_worker._get_images_dir() to check the kolla
package directory first so both editable and regular installs work.
Change-Id: I4fc452c03740ef22ebbb4e8d0d86c4f4783e2d09
Signed-off-by: Michal Nasiadka <mnasiadka@gmail.com>
|
|
The build-date label was rendered into a LABEL instruction on line
3 of every Dockerfile template, immediately after FROM.
Because Docker invalidates the cache for a changed instruction and
all subsequent layers, a date change caused a full rebuild of every
image on each new calendar day, even when nothing else had changed.
Fix this by removing build-date from the Dockerfile templates and
applying it as a --label argument at build time via image.labels.
Labels passed on the build command line are not part of the
Dockerfile cache key, so the date is still recorded in the image
metadata without affecting layer caching.
Change-Id: I8ef0d85bab1a999a4d09f7cbf0c088ca00fdcf77
Signed-off-by: Michal Nasiadka <mnasiadka@gmail.com>
|
|
Currently we are enabling repos irrespective of rpm/deb
statement blocks and not failing in the method used
if a repository is missing.
Downside is - we might be producing one more layer,
but it probably is a more logical approach.
This is required for adding rpm support in repos.yaml
that will also include failing on missing repositories.
Change-Id: I5479c5c935760f00fe4cd307366f261deee3199f
Signed-off-by: Michal Nasiadka <mnasiadka@gmail.com>
|
|
RDO has staffing problems and has decided to only
support SLURP releases - let's drop reliance on that
repository.
Change-Id: I56f86a1c5262e3ee2bcbb52cae53c3b2655ed87f
Signed-off-by: Michal Nasiadka <mnasiadka@gmail.com>
|
|
This patch adds a way to patch files in a Docker
image built by Kolla. This is very useful for several
reasons, specifically:
- Custom modifications
- The stable branch of some library has a fix but no pip
package has been released
- Eliminates the need to package your own pip packages
- Eliminates the need to invent your own versioning to prevent
upstream versioning
- Eliminates the need to manage a pip server
- In other words, it eliminates the need to get a wheel into
the image and install it manually using any method not
previously mentioned
It is also highly desirable because, although Kolla can replace
the source for a service with a custom URL for a tarball or its
own Git repo, it cannot do this for dependencies pulled from pip.
I would also like to point out that this is a feature with its own
code path and works only if the user "inserts" a patch into the folder
patches/docker-image/something.patch and creates an analogous series
file for patch source code.
Simply said, this code will never interfere with the upstream build process
since this feature is not intended for use in upstream.
It is rather meant for downstream users who know what they are doing.
Now they just have an option to patch their images.
Everything works on all layers of the Docker image and stores a report
of applied patches which can then be seen in /etc.
This mechanism is similar as debian patch quilt.
Change-Id: I61d0790c5d4d070b7ea9e8c99c0a76ff5d22bf9d
|
|
Fixes a hypothetical security issue related to privilege escalation via
rootwrap/privsep. A potential vulnerable service could previously allow
writes to its rootwrap/privsep config and thus allow for more commands
to be run with root privileges via rootwrap/privsep. For a succesful
attack, this would also require the service to allow to run arbitrary
commands via rootwrap/privsep. Thus far, no such vulnerabilities have
been reported and thus this fix is simply strengthening the container
images against such an issue in the future.
Change-Id: I92c81c77e6a16570a108cde8031f7977930fb02a
Closes-Bug: #1874298
|
|
As we have one type of images now some RUN calls could be merged so we
will have less layers in resulting images.
Change-Id: I5178c58fbd8c65efe825dc249c0f1368ef0fe8e0
|
|
Explicitly set the permissions on the kolla-toolbox kolla_extend_start
file. Also, since all extend_start files are sourced rather than
executed, the executable bits are now cleared throughout the project.
Change-Id: I5c2deb4a2e33575d57c852089f856a9acc6818d0
|
|
Big patch drops all mentions of binary images support. Suggestions are
welcome how to split it into parts or handle better.
Change-Id: I5d5a46c6ce7734ceb8b844e17b43e359d7cac6e3
|
|
With RDO use we did not disabled some repositories. This patch disable
them and enable where needed.
Change-Id: Ia9d537fe9c1ad54789d2bfb4027254fbb3defe7e
|
|
Change-Id: I9692dda817ef134d647247431565e1b58cf9da41
|
|
This change updates the docker files to use base_package_type instead
of doing specific distro checks for the rhel/deb generic cases. The
base_distro is still available and is used when a specific distro needs
a customization but if the differences are purely rpm vs deb, then the
base_package_type can be used.
Change-Id: I8d720bb185df65a0178061ccf20b1ab2265da2c5
|
|
centos based images have wrong label info,
these changes fix own image's name and build-date.
Change-Id: I1d13f8f386c8db12b5fbe5f8ecbbf9e3fbb4ba1c
Closes-Bug: #1680341
|
|
Use LABEL instruction instead of MAINTAINER (deprecated) instruc-
tion as suggested by Docker's official dockerfile guide.
docs.docker.com/engine/reference/builder/#maintainer-deprecated
Closes-Bug: #1683652
Change-Id: Ie87a1ddf31aefcd0b623fd2837d78de420e76898
|
|
Debian support is not maintained in Kolla so it got a bit behind Ubuntu
one. This changeset enables Debian for all images. Jessie (even with
backports) may be too old for some images though.
Also unify distro check to ['debian', 'ubuntu'] to keep alphabetical order
like it is done for RPM distributions.
Partially-Implements: blueprint multiarch-and-arm64-containers
Change-Id: I056233fbfa277e0e2360c07c3f80d9558c554357
|
|
1. Enable customization of pip packages in source
branch of most images
2. All pip packages install uniformly through
install-pip macro, user can easily customize his
own pip command (For example using a mirror)
Co-Authored-By: Mauricio Lima <mauriciolimab@gmail.com>
Change-Id: If09582039f690fa4136e8f33200d5da15e092da7
|
|
source and binary alll need sudoers
Change-Id: I190a0b8cbbb05460e1ff80e0e7a30d05c3392640
|
|
This centralizes all user and group creation into a single source. This
will fix any current and furture uid/gid mismatches (such as with
nova-libvirt).
In the process, we also unify users between the distros in a standard
way. The users in the following containers change from thier defaults:
Ubuntu: _chrony user is now chrony
Ubuntu: memcache user is now memcached
All: qemu user is used for ownership and socket permissions
All uid and gid numbers are customizable via kolla-build.conf
Co-Authored-By: Kris Lindgren <klindgren@godaddy.com>
Change-Id: I120f26ab0683dc87d69727c3df8d4707e52a4543
Partially-Implements: blueprint static-uid-gid
|
|
Change-Id: I39aa88489f744f779150695f3f55ef80d42e1c61
Closes-Bug: #1653247
|
|
Closes-bug: #1616387
Change-Id: Id97f88b9baa3d48d33ce120962450a374282d044
|
|
Change needed to add header blocks to all Dockerfiles, similar to the
base.
Use case is to easily run something before packages are installed, e.g.
to COPY a local rpm in that can be added to the package list.
Change-Id: I1bbfdf0b762da0a392aa8bf47781315b45377bee
Closes-Bug: 1618969
|
|
Is a best practice in Unix/Linux scripts to use dots
instead of source command.
Using dots will avoid issues with non BASH shells
TrivialFix
Change-Id: Ie6480a1954f853f79faffa093452715ebd9f7d90
Signed-off-by: Eduardo Gonzalez <dabarren@gmail.com>
|
|
Currently if the install_packages macro is run with an empty
package list, it will add a yum or apt-get command with no
packages listed.
This bug fix aims to omit this line when no packages have
been given, or, the operator wants to use the "_override" /
"_remove" functionality to disable all packages being
installed in a Dockerfile.
Co-Authored-By: Paul Bourke <paul.bourke@oracle.com>
Change-Id: Ifaaaebfccc3adb0f2f68a35ac08e59378bc87fdb
Closes-bug: 1612446
|
|
Change-Id: I3383f206b66d1af39273ec0564a53f1f7a592527
Partially-implements: blueprint third-party-plugin-support
|
|
Change-Id: I4b109f7fdc3b8e49defed26979b04ca158842e98
Closes-Bug: #1598423
|
|
Change-Id: If4be00b937e14ec93443dcb7249cf17099d57cbe
Closes-Bug: #1569417
|
|
manila-share is present in manila-common package.
Change-Id: Ida231c0468b6ee3bb6a76fb9f475724abf95c1cb
Closes-Bug:#1553199
|
|
Change-Id: I8277ad668f46d327807a01ceb96c2abb417f0e3d
Closes-Bug:#1556271
|
|
Partially implements: blueprint heka
Change-Id: I3fadbf29f5219a293b8b1114d192e27888be2c59
|
|
All manila rootwrap calls are failing in
source build because lack of sudoer setting
Change-Id: I301fd7b6ff3721328828c2e0384f7e320146731b
Partially-Implements: blueprint enable-manila-containers
|
|
Change-Id: I1ad667487ce1e0fbae340bed2233aacd4f895bce
Parially-Implements: blueprint enable-manila-containers
|
|
Because we use rootwrap in a venv we need to update the exec_dirs.
Without doing this *some* commands will break for various reasons that
I won't get into in this review in detail. But this is required for
neutron-l3-agent in a container + drop-root.
Change-Id: I1a09f7188fdd501b7ce251d9f2fb0e5b10222142
Partially-Implements: blueprint drop-root
|
|
Change-Id: I5b40de377ac0323d3253f58b49318d57da347d11
Parially-Implements: blueprint enable-manila-containers
|
|
Change-Id: Iea77fa26c17e120b5c0487eca8a76972281ee99a
Partially-Implements: blueprint enable-manila-containers
|
|
Support manila as container. First step only supports
building from source.
Change-Id: I60bb67536c9afdb9f0532b3cdc2c400a68608003
Partially-Implements: blueprint enable-manila-containers
|