Age | Commit message (Collapse) | Author | Files | Lines |
|
A cleanup stage is also added to clean testing
artifacts.
Change-Id: I2255a0e90a0f609bd927d3d44dd05b5847fbdfb0
Issue-ID: OOM-1926
Signed-off-by: Bartek Grzybowski <b.grzybowski@partner.samsung.com>
|
|
|
|
|
|
|
|
If host networking is not used for dns container,
dns wont work for other infra containers. This is
unintentionally removed in
https://gerrit.onap.org/r/#/c/oom/offline-installer/+/87422/
Change-Id: I07ce6874e7baedadbdbbcf0913f1d683e19f93cd
Issue-ID: OOM-1866
Signed-off-by: Michal Ptacek <m.ptacek@partner.samsung.com>
|
|
|
|
Change-Id: Idddc421d33698c08b19cc63f4e203d0964341462
Issue-ID: OOM-1913
Signed-off-by: Bartek Grzybowski <b.grzybowski@partner.samsung.com>
|
|
Introduce full role level test coverage for
'nginx' role.
Change-Id: I7c634775433a2336e43cdfc73d4698d5e2933f0e
Issue-ID: OOM-1913
Signed-off-by: Bartek Grzybowski <b.grzybowski@partner.samsung.com>
|
|
This commit adds molecule tests for kubectl role.
That includes verification for kubectl to be installed and runnable.
Change-Id: I4cb0679ed6826555a6d1f6cff4542ae52869f404
Issue-ID: OOM-1908
Signed-off-by: Michal Zegan <m.zegan@samsung.com>
|
|
This adds molecule tests for the helm role.
Change-Id: I5be27992062ca9039c141318bc54ce85ce35526b
Issue-ID: OOM-1909
Signed-off-by: Michal Zegan <m.zegan@samsung.com>
|
|
|
|
This commit removes bindmounting
of host directories to containers in play-resources tests.
It seems that it is unneeded, and it is enough to just create anonymous volumes
instead of cluttering host filesystem.
This is one part of a preparation for removing the cleanup-directories
role that won't be needed anymore.
Note: similar changes were made to the nfs
role tests, for the same reasons, but
cleanup-directories is not used there
Change-Id: Ib33543427d72b4bc7ede532e5eb9a5023931f1fe
Issue-ID: OOM-1902
Signed-off-by: Michal Zegan <m.zegan@samsung.com>
|
|
As of Ansible 2.5 'loop' keyword is a recommended way
for looping statements instead of 'with_dict'.
Change-Id: I19e00cc27e13955e2056840fd4ce99841bc824ad
Issue-ID: OOM-1887
Signed-off-by: Bartek Grzybowski <b.grzybowski@partner.samsung.com>
|
|
In Dublin:
First project has taken ownership of own helm charts, oom repository must be obtained with subrepositories.
By default no applications are enabled, introducing a way to submit more override files in to installer
Issue-ID: OOM-1861
Signed-off-by: root <m.pilat@partner.samsung.com>
Change-Id: I1724de6b87bd0e9d17df4d0f55615a3f1c439d40
|
|
This change adds molecule tests for the rke role, and
modifies the rke role itself to be more idempotent/to
pass linter.
Note that this molecule test case uses a separate
role to install docker in containers, that runs
docker daemon inside of them instead of using host docker.
Issue-ID: OOM-1778
Change-Id: I875f3ff2ab961e5428acee5a02287a8d2d6e9969
Signed-off-by: Michal Zegan <m.zegan@samsung.com>
|
|
|
|
|
|
|
|
This commit supplements 'Molecule test for
infrastructure playbook' with role level tests.
Change-Id: I14cdeedf1c80c0ee06bacf0a391ee1d7ccd6df55
Issue-ID: OOM-1841
Signed-off-by: Bartek Grzybowski <b.grzybowski@partner.samsung.com>
|
|
This change modifies molecule tests of rancher role to
use prepare-docker-dind.
Also:
- it modifies prepare-common role to retrieve cluster_ip of the actual
node in question,
- it fixes the rancher role itself to work at all by removing auto_remove.
It has been tested by running molecule converge.
Change-Id: I7af4e9998feedd5f8fc3df8215e03c1b0485529c
Issue-ID: OOM-1864
Signed-off-by: Michal Zegan <m.zegan@samsung.com>
|
|
* changes:
Molecule test for infrastructure playbook.
Add the prepare-package-repository role
Add prepare-nexus role
Add prepare-nginx role
Add the prepare-dns role
Add prepare-vncserver role
Allow unsafe writes when modifying resolv.conf
Add nexus_come_up_wait_retries variable to nexus role
Remove docker restart condition from docker role
|
|
The package-repository role edits /etc/resolv.conf.
Ansible modifies files atomically to prevent data loss on crash, but that
does not work in docker containers because /etc/resolv.conf is
bind-mounted.
Allow ansible to fall back to non atomic writes in such cases.
This makes it unnecessary to hack around this.
Change-Id: I066205322de5cc06569e6ea3aa38885d3834f9cd
Issue-ID: OOM-1816
Signed-off-by: Michal Zegan <m.zegan@samsung.com>
|
|
This variable is used to set the number
of retries when waiting for nexus to be running.
It may occassionally be useful to increase it on slower
test environments.
Change-Id: Id34106dd8f9a8b2b1949b80adb22fbec3df71e23
Issue-ID: OOM-1816
Signed-off-by: Samuli Silvius <s.silvius@partner.samsung.com>
Signed-off-by: Michal Zegan <m.zegan@samsung.com>
|
|
Docker role doesn't trigger docker restart when run in docker container
(e.g. molecule). Remove the condition, as in dind case
docker in container works same way as docker out of container.
Change-Id: I8c80a09baccd76c78f2cbb8e474044351fd9b3d3
Issue-ID: OOM-1816
Signed-off-by: Michal Zegan <m.zegan@samsung.com>
|
|
Change-Id: Ie0b4e812acf6fba4e4b8f082533ab2f998456095
Issue-ID: OOM-1877
Signed-off-by: Petr Ospalý <p.ospaly@partner.samsung.com>
|
|
This modifies the docker role to use dind.
Note that it now uses pepare-docker-dind role, but
*disables* start of docker.
Change-Id: I55ac8d1f69531ba7496c34f990818b894a7d02a5
Issue-ID: OOM-1864
Signed-off-by: Michal Zegan <m.zegan@samsung.com>
|
|
Currently molecule tests for certificates role call the
docker role just to make one handler available.
It actually breaks best practices of unit testing,
the docker role shouldn't be called nor tested in
unit test of certificates role.
This commit makes certificates role independent by just
creating fake handlers.
Change-Id: Iec7934c7c1f4e9b37ff523e1d86a3beaaa88bdcf
Issue-ID: OOM-1864
Signed-off-by: Michal Zegan <m.zegan@samsung.com>
|
|
Two problems are fixed in this commit. Firstly handler
is not triggered so dnsmasq is not started. Secondly
dns is not working due to loop in dns settings.
Change-Id: If6874c99dcc42ef63da167abe8d1f7cdd8e0fcb1
Issue-ID: OOM-1866
Signed-off-by: Michal Ptacek <m.ptacek@partner.samsung.com>
|
|
* changes:
Upgrade docker version to 18.09.5
Update of docker for RHEL
|
|
Either run kubectl proxy, but you must setup routes or kubectl on some
node (only infra has kubectl by default). Another option is connecting
onto the node where the dashboard is exposed via NodePort. Instructions
are here:
https://github.com/kubernetes/dashboard/wiki/Accessing-dashboard
(1.7.X and above)
Change-Id: I5a28f458c0ae949278b1e6eaf6eb063d09f93fc0
Issue-ID: OOM-1778
Signed-off-by: Petr Ospalý <p.ospaly@partner.samsung.com>
Signed-off-by: Michal Zegan <m.zegan@samsung.com>
|
|
* changes:
Add setup for kubectl and helm
Add support for RKE kubernetes implementation
|
|
This patch upgrades docker version for
Molecule tests to 18.09.5; it also sanitizes
tasks in 'docker' role to supplement this
change: https://gerrit.onap.org/r/c/87761/
Change-Id: I14beffafb129b24e7c15f8da45f1a25e27ccae9c
Issue-ID: OOM-1852
Signed-off-by: Bartek Grzybowski <b.grzybowski@partner.samsung.com>
|
|
the kubeconfig is copied from the cluster config directory to ~/.kube/config.
Also, the playbook runs kubectl and helm roles to install tools.
Change-Id: I29ee98f6502bbb7923b1ae6812dd642ca206dbce
Issue-ID: OOM-1778
Signed-off-by: Petr Ospalý <p.ospaly@partner.samsung.com>
Signed-off-by: Michal Zegan <m.zegan@samsung.com>
|
|
Added a new playbook rke.yml and role rke which uses rancher RKE:
https://github.com/rancher/rke
It's an implementation of the kubernetes from rancher.com folks
and it is an alternative to the official kubernetes tool: kubeadm.
NOTE: Rancher has a notion of a 'control plane' which cannot run with
etcd on all nodes in a multi-node setup. Control-plane node is the
first kubernetes node from the inventory (as of now).
Change-Id: I0bf669442a5183efa20d44fb1cac823e7ce54348
Issue-ID: OOM-1778
Signed-off-by: Petr Ospalý <p.ospaly@partner.samsung.com>
Signed-off-by: Michal Zegan <m.zegan@samsung.com>
|
|
Issue-ID: OOM-1812
Change-Id: Ifb6f5a10afb4014b20be77a4141371e561d346ce
Signed-off-by: Samuli Silvius <s.silvius@partner.samsung.com>
|
|
Docker_container module was called twice: in a taks
'Start dns server container' and then as a handler
but there was a divergent set o parameters for the
module in them. This had the efect of the container
being started by the means of the task itself but
immediately after it was killed and started again
by the handler because it had different set of
arguments for the module. The fix is to include the
handler itself as a task so the whole run is idempotent.
Change-Id: I9078f9dfe5894d94d4b6a9e44411f034508992e7
Issue-ID: OOM-1845
Signed-off-by: Bartek Grzybowski <b.grzybowski@partner.samsung.com>
|
|
|
|
This commit reverts removal of 'ignore_unreachable'
option in 'cleanup.yml' for 'nfs' role tests in
96a4baf. It's needed for cleanup stage to succeed
if called prior to 'create' stage (default 'test'
matrix behaviour)
Change-Id: I9177fb950786f05a55c45a0089fb3632503f14f5
Issue-ID: OOM-1809
Signed-off-by: Bartek Grzybowski <b.grzybowski@partner.samsung.com>
|
|
Molecule role tests that use pre build images had
superfluous 'container' env variable declaration.
It's not needed as this env is already set in
Dockerfiles, for both: pre-built and custom built
images.
Change-Id: Iee9cc38919da894ee6cc6a2a5f1acc60d76865de
Issue-ID: OOM-1841
Signed-off-by: Bartek Grzybowski <b.grzybowski@partner.samsung.com>
|
|
Rancher role test do remove rancher/agent and rancher/server
containers at 'cleanup' stage but Rancher Agent spawned
containers are left orphaned. This patch adds tasks to
remove them.
This patch also adds custom ansible module for listing
docker containers (as of ansible 2.7.10 no upstream module
provides that feature, even new docker_container_info info
module in ansible's devel branch doesn't have it)
Change-Id: I6325dc81063b55b70136280273f8f6138c7a0375
Issue-ID: OOM-1811
Signed-off-by: Bartek Grzybowski <b.grzybowski@partner.samsung.com>
|
|
* changes:
Add support for resetting the admin password
Add support for rancher authentication
Refactor rancher1_api module
Add support for rancher 1.6 API
|
|
The module rancher1_api can now reset an admin password of the default
admin account (account_id: '1a1') and keep local authentication enabled.
By default the authentication is not enabled, because the ansible
module rancher_k8s_environment.py is not idempotent and it would need to
be rewritten.
Change-Id: Ib432537651b91216c32438ec1233dba3602e3faf
Issue-ID: OOM-1734
Signed-off-by: Petr Ospalý <p.ospaly@partner.samsung.com>
|
|
This commit adds a new mode to the rancher1_api module, which enables
the rancher local authentication (username/password).
There is an already predefined rancher admin user called 'admin' and
that is the account, which this mode modifies. Due to the complex
API and the fact that rancher 1.6 is soon to be obsoleted, this module
is going the simpler route and it is just editing this default admin
account instead of creating a completely arbitrary username/password
credentials. For that reason is using the 'account_id', which is unique
for all accounts and the default admin account of rancher has '1a1'.
As of now this module cannot handle changed password once the auth. is
enabled.
Change-Id: Iea8923c71bdb82267c966a00d62f0f43eb5adb76
Issue-ID: OOM-1734
Signed-off-by: Petr Ospalý <p.ospaly@partner.samsung.com>
|
|
|
|
Moved 'file' package installation from Molecule's
'prepare' stage to Dockerfile to shorten the test
invocation.
Change-Id: Icbb805c54a4fd42003d0cde037fa56208bd6008a
Issue-ID: OOM-1809
Signed-off-by: Bartek Grzybowski <b.grzybowski@partner.samsung.com>
|
|
Cleaning up of nfs mounts should only happen
on clients, not on server itself.
Explicit declaration of 'cleanup' stage playbook
not needed as it's in the default location.
Change-Id: I54e00220048f01160c15fc9631ae62586596114c
Issue-ID: OOM-1809
Signed-off-by: Bartek Grzybowski <b.grzybowski@partner.samsung.com>
|
|
This rewrite enables to add easier more features supported by the rancher
API. The initial idea of a simple get and set through the JSON REST API
is not feasible. To achieve something with the API one may have to setup
more options on different URLs and in a particular order. For this reason
the module comes with the mechanism of "modes", which is a wrapper around
some feature in the rancher and which can require multiple steps to do.
Rancher1_api module could also support "raw" mode where the user will not
be limited by a few implemented modes in the module, but he can craft all
requests by hand - but due to the fact that such thing can be done easily
with just curl command and ansible shell module, there is no benefit in
doing so. Especially when rancher 1.6 is already obsoleted within the ONAP.
The useful value of this module is to give the user a simple means to set
something in the rancher and hide all the boilerplate from him via a mode.
- Original logic was rewritten to utilize the "mode" mechanism.
- New module structure is also easier to test outside of ansible.
Issue-ID: OOM-1681
Change-Id: I0e7932199df9ec1acd80af545060199867ad17fa
Signed-off-by: Petr Ospalý <p.ospaly@partner.samsung.com>
|
|
- New ansible module to handle rancher API
- Setting up of the cattle db and log related options
to achieve lower space usage.
As of this moment it does what was intended:
- it setups new values for db and log related options
- it can be used for any other setting options
Change-Id: I25048469df0cb035cc6eac39740210cdfa175ada
Issue-ID: OOM-1681
Signed-off-by: Petr Ospalý <p.ospaly@partner.samsung.com>
|
|
|
|
This commit improves the nfs role by
using cluster_ip variable instead of ansible_host for getting
nfs client/server address, and
using /etc/exports.d instead of /etc/exports for
exporting nfs filesystem.
Also, this commit fixes other
issues in molecule found along the way.
Change-Id: Ibd4ea5d4fd937116e2dc9e1b6f85c3e019f62e17
Issue-ID: OOM-1770
Signed-off-by: Michal Zegan <m.zegan@samsung.com>
|