| Age | Commit message (Collapse) | Author | Files | Lines |
|
When an image did not provide a build script, due to a broken link for
instance, the build-all-docker-images script was wrongly marking the
image as "up-to-date", and the functional gate was reporting success.
This commit restores the expected behavior, and mark the images with
non-existent build scripts as failures.
Change-Id: Ifedc8d9c312925fe5aa953ed8f8fdae02b756906
Closes-Bug: #1460959
|
|
Change-Id: I5469e6e60b9e20ee98c9149697f663e94e58077d
Partially-Implements: blueprint bashate-gate
|
|
Change-Id: I62dbd2c3e0e6f66b1be5777bb8bbda4e43b2b5db
Partially-Implements: blueprint bashate-gate
|
|
1. Move setup_docker.sh to tools directory
2. Make a setup_gate.sh that installs necessary packages including
docker and starts docker.
3. Add logging output.
4. Add default test timeout of 2 hours.
5. Add user to the docker group before running test cases.
6. Run image build as dockerroot group.
This patch has to be one commit to fix the gate in one go.
Co-Authored-By: Steven Dake <stdake@cisco.com>
Change-Id: I83f3cdb1dabf0dfface589c581cb22c155467acc
|
|
Instead recommend developer run build within the docker group.
This script is non-functional as the root user.
Change-Id: Ib70ab55e5e9bc2e7b8639d5d5ffc32a2b8795058
|
|
When we're not providing a tag, the build script will use one based on
the short SHA of the tip of the current git branch.
Make the `build-all-docker-images` script always provide a tag to
`build-docker-image` in order to prevent issues with Docker not being
able to find the base image when changing branches during an ongoing
build.
Closes-Bug: #1458739
Change-Id: Id6925b792f46d14b20a7dbebfaf0d351ddec6538
|
|
|
|
|
|
|
|
(neutron client version: 2.5.0)
The option is --dns-nameservers not --dns_namservers and the
subnet-create subcommand expects the NETWORK and CIDR to be adjacent
args.
Change-Id: I4f776632902fd001c38ee5c8c87a213883ca87be
|
|
As a restructure, nothing is changed from the original behaviour and
naming despite the file structure changing. The symlinks to build had
to be updated generating lots of "deleted" and "new_file".
The new structure is:
docker/${base_distro}/${type}/${container}
base_distro == centos, ubuntu, fedora, etc
type == source, binary, rdo
type rdo is a symlink to binary for backwards compatibility
Two new flags are added to the build-all script to support the ability
to support different base distros and a flag to support binary or source
containers.
There are several added folders that are empty to hold the directory
structure for future containers of these types.
To use a prefix other than centos-rdo- you can set PREFIX in the toplevel
directory .buildconf file
Change-Id: Ifc7bac0d827470f506c8b5c004a833da9ce13b90
|
|
|
|
This adds functional tests to tox, along with a number of prerequisite
steps. Since this is primarily a bash project, pbr usage has been avoided.
Tests are to be written in python and put in the tests directory.
Running tox -e setupenv will execute a setup script
(tests/setup_docker.sh) to ensure the running instance of docker meets
the minimum version requirement.
Running tox -e images will execute the image building script
(tools/build-all-docker-images) and will parse the resulting output to report
failures.
Running tox -e startenv will generate the environment file, run
"tools/kolla start", and run first time initialization (eventually).
Running tox -e functional is for actually testing the deployed OpenStack
environment via a series of tests utilizing the client APIs.
Change-Id: Iff6dfdca43f0c44d471e7540a7836e56a0de4507
|
|
|
|
Change the sed seperator to accept a namespace with a private registry
and add an option to specify the registry. Below is an example for a
registry located at 192.168.0.10:5000.
$ ./build-all-docker-images -r 192.168.0.10:5000
The final image name would be something like this:
'192.168.0.10:5000/kollaglue/centos-rdo-base'
All build, push, and pull commands will still function the same.
Change-Id: Ic0ff96bb8119ddfab15b99e9f1e21cfe8d321dab
|
|
Previously, vnc was not completely working properly because
nova.conf config flags were not properly implemented in t he correct
containers.
This patch moves the nova.conf config flags for vnc to the proper
containers. Additionally, the patch modifies the proxy client ip
param to be consistent with the param within nova.conf and adds
all the novnc params to docs.
Change-Id: I1944baca64ccd113bbca582784b08f90dc0eff69
|
|
|
|
|
|
The --router:external option is a boolean flag that doesn't take a value
(neutron client version 2.5.0). Passing True results in the following
error:
Invalid values_specs True
This patch simply removes the 'True' value.
Change-Id: If4b5305f37d50c14c767393eb21223b0e0b59494
|
|
Convenience commands for restarting all containers and destroying all
containers
Change-Id: I9c4759dbe699bd4b609cefa694ecc897355ff60a
|
|
Change-Id: I418583af2d2795f4b63ded2aa3e2cbedd70993c8
|
|
Make a Magnum demonstration that creates a baymodel and bay. Just
run demos/magnum/launch to make it work. This depends on Neutron.
Borrow the v1beta3 redis example from the examples repository in
Kubernetes.
Change-Id: I448a5890bfe0c1675914ae7dbd02fad03f4b1eeb
|
|
This adds Magnum Containers as a Service to Kolla.
Change-Id: I572cd62845515f42a6a16291d2bd87ad8d972e24
Implements-blueprint: container-magnum
|
|
Also support --pull flag to pull the latest version of the base
Change-Id: I3205d6e799772138edaf2b0666353c19dfbf52f0
|
|
|
|
Previously cleanup-containers suffered from the following
limitations:
1. The docker rm command would fail if all the containers were not
stopped.
2. The script would throw an error if no containers existed.
3. Artifacts such as veths, bridges, etc.. would remain on the host
after containers were removed. These artifacts should be removed
if the containers are removed.
Change-Id: I5eadb23fec9e34125a32f6c3e39c241d97932910
|
|
Use kilo upstream RPM packages
Partially-implements: blueprint port-kilo
Change-Id: Ia0b4db5dc8e40c01184871e77458dc718c3c46cd
|
|
All VNC-specific configuration bits are moved from nova-api-base to
nova-compute and the new nova-novncproxy and nova-consoleauth
containers. The compose specification has been updated to reflect this.
A future patch will make the desired backend configurable (spice/vnc).
Change-Id: I8b64abf125fd1ecd7d3463515e3717dd9bd60420
Blueprint: novnc-container
|
|
This represents the Juno release of the images.
Partially-implements: blueprint port-kilo
Change-Id: I2837ec25a2aa093b7fff7519f3ed608ec93736bf
|
|
Add Heat-api-cfn for wait conditions
Magnum needs wait conditions in order to operate properly. heat-api-cfn
is where wait conditions connect.
Change-Id: I56742dd96e4607dc0042d0cef4e18c922f9ed736
Implements: blueprint container-heat-api-cfn
|
|
Introduce a new --no-use-released-parent option to `build-docker-image`
that allows specifying whether the parent image uses the same tag as
the one that's currently build or the release tag.
Default is to use the release tag for parent image, however
`build-all-docker-images` script forces it to be the current tag.
Change to Dockerfile is coming next.
Change-Id: Ief11c9abb722c181ca85f38ee11800dbf4332a5f
|
|
The cirros image download and creation now happens in
tools/init-runonce.
Change-Id: Icb83526c1fbe42b96957dd39ad9e46a1cf450471
|
|
The network name must be next to the CIDR otherwise `neutron
subnet-create` gets confused.
Change-Id: Ie8b19d3824e5e25f80c61f0794595813c80b98da
|
|
|
|
|
|
|
|
Add "-a" to preserve file attributes so that unnecessary Docker image
build can be avoided. Otherwise, build-all-docker-images will generate
additional image layer due to timestamp change.
Change-Id: I893fa0637937c521198f50b1bba203a27c9f00a3
Closes-Bug: 1449287
|
|
Fix issues reported in https://review.openstack.org/#/c/177681/3, and
simplify detection of cirros images in glance.
Only add developer's SSH key to nova if ~/.ssh/id_rsa.pub exists and is
readable.
Change-Id: I5936c6db399d63902d460b0d3c959619073cc3cc
|
|
Configure the network for use with Neutron.
Co-Authored-By: Daneyon Hansen (danehans@cisco.com)
Change-Id: Iab0496cf40683ca29cbe38066ae446aa8c6c4c36
|
|
Configure nova networking and nova quotas to allow 40 m1.small
instances to run. Checks for keystone credentials set. May only
be run once.
Change-Id: I36e01d42ec283651babbb59268452a65e644626e
|
|
Unify the shell to bash in all Kolla scripts.
Change-Id: Ib9591b2f8f344eb88455c5e9b7ecf2164fb5960a
Implements: blueprint use-bash-shell
|
|
Should check compse cmd excute result
can't let it go.
Change-Id: Ifd20e3b7f4ae385f7d45ab119a8850f44f254bac
Closes-Bug:#1447959
|
|
The kolla script runs its process on rabbit, not rabbitmq. This
results in the rabbitmq compose scirpt not being executed.
Change-Id: I80f8d537c36b6cd598cdcf87f77df418db8ed035
|
|
|
|
|
|
This is supposed to prevent proliferation of scripts in "tools"
directory and reduce code duplication.
This first patch replaces tools/{start,stop,pull,status} scripts and
more commands are to be added in the future.
Change-Id: I3d99cf5033be8631d8e6f32a4c34d3b5ffcae7e8
|
|
|
|
NEUTRON_SHARED_SECRET was being generated incorrectly. It had a colon
at the front, which causes nova not to operate.
Change-Id: Id0d9390fa60b3d946513098fed394aaba3a53c70
|
|
Previously, only start/stop scripts existed. The status script
runs docker-compose ps for each container so users can
get a quick snapshot of their Kolla environment.
Change-Id: I3204f3e7404ad906d5330cf67d091fc1f06d3030
|
|
Previously, Neutron was not included in the start/stop scripts.
This patch reads the NETWORK_MANAGER value from openstack.env
to start/stop either nova or neutron networking.
Change-Id: I85b31396c2b816b89f33206443c3d60cb4b7914b
|