Add two tests with the following steps:
Deploy environment with enabled DMZ network for API.
Scenario:
1. Revert snapshot with ready master node
2. Create new environment
3. Run network verification
4. Deploy the environment
5. Run network verification
6. Run OSTF
7. Reboot cluster nodes
8. Run OSTF
9. Create environment snapshot deploy_env_with_public_api
Check that security rules are properly applied for DMZ network
Scenario:
1. Revert snapshot from previous test
2. Run instance
3. Try to access horizon from instance
4. Remove instance
Implements: blueprint test-separate-public-floating
Change-Id: I70474b5cab324aa4f4a042127d4e6961c95010bf
If we using default_devops_templates we should use relative path
and get_devops_config function help's with this
Change-Id: I2bfc9ddaaa28d416c26a7cd659ee117fe9f124f9
Fix for correct work of disabling mos-updates|security repos if
PATCHING_DISABLE_UPDATES=True for fuel-qa/master branch
Change-Id: Ie39caa26d93c49fcf78a134253dddb125acf621c
1. test_os_upgrade is moved under test_os_upgrade
2. main code is moved to separate place for excluding copy-paste
in new tests.
3. Fixed logging due to unreadable output
4. Switch to check_call -> remove pain of cherry-picking
5. Use octane-cleanup
Closes-bug: #1612236
Change-Id: Ida7f5901f070a9ef507ce6027fd2618b8617d89f
(cherry picked from commit a9f7dd2)
Now in latest devops it's possible to insert cdrom tray without
virsh magic - now it's time to remove it.
Change-Id: I762cc449c4715e360d794beec872b1fd79273215
Depends-On:I90b4d4884914e0a5be914c1429dffef33a33e922
1) Need to use fuel-devops >=3.0.0
2) Need to use centos_master.yaml devops template
3) export CENTOS_MASTER=True environment variable
4) provide path to CENTOS_CLOUD_IMAGE_PATH, FUEL_RELEASE_PATH and
EXTRA_DEB_REPOS
Change-Id: I1542c2238abc364713f02e4bca6ec7646883bf78
Closes-Bug: #1592419
This test is useful for preparing baremetal environment
for further manual testing.
Related blueprint run-fuel-qa-on-baremetal
Change-Id: Id3002ffc147e284a5d349d13b43d9a741d7fce04
In case SEPARATE_SERVICE_HAPROXY_PLUGIN_PATH is set without
MULTIPLE_NETWORKS we got local variable 'load_yaml' referenced before assignment
Since this variable is added for all plugin tests it breaks them
Partial-bug: 1589914
Change-Id: I8e2dc6f6074db6c3d45d73a102b5d9f9e9671f25
- Add test with external load balancer. Controllers are
from different racks and haproxy is from rack-3
- Separate devops config with appropriate networks
assigned to nodes is used
- Local repos for cluster are used because public networks
are routed without internet connection
- OSTF isn't running because it's not implied to use
separate haproxy
Closes-Bug: #1583530
Change-Id: I0d3647c8eb13159c27e64ddf5925467f451b610c
SSHManager.update_connection: add possibility to open new connection with
parameters, if no connection presents.
This testgroup should be run with this parameters:
export MULTIPATH=True
export SLAVE_MULTIPATH_DISKS_COUNT=2
export MULTIPATH_TEMPLATE=system_test/tests_templates/tests_configs/multipath_3_nodes.yaml
Implements: blueprint test-multipath-devices
Change-Id: I73ffc4338b59c8b44eabe27b2d759b7fbbba19a4
Add test for testing Separate master node deployment
on pre-provisioned Centos 7 server
Change-Id: I9b9d40a59d24f579502a38dfc9b8c142bc219a06
Closes-Bug: #1575650
Move some actions executing after master node boostrapping to
TestBasics action to using them in separate_master node deployment
Implements blueprint: fuelwebclient-refactoring
Change-Id: Idb8954c71861840124a805e8982a2fe63c94376f
1) introduce additional pacemaker helper to get correct resource
name for multi-state resources
2) alter strength tests to use the correct name
Change-Id: Ibe97e2d7cadf475354e1ed77b41deff6faf68151
Closes-bug: #1551708
Refactor tests for multiple cluster networks feature so
all tests use fuel-devops environment which has 3 nodegroups,
shared L2 network and it's defined using template. Also allow
tests to configure networks for more than 1 custom nodegroup.
Add aditional check for shared storage network (Ceph health).
Create new test case for adding new nodegroup to operational
environment.
Co-Authored-By: Artem Panchenko <apanchenko@mirantis.com>
Closes-Bug: #1521316
Implements blueprint test-nodegroup-add
Implements blueprint test-nodegroups-share-networks
Change-Id: I066248a7b96a6f16b1e24763892234fe48803983
- move self.env from attribute to a property in TestBasic class.
- save devops template to TestBasic attribute in order to
prevent environment model initialization without config
Change-Id: Ifd3ae4db66f37992ff8f956491882915f8fc1ec3
Closes-Bug:#1528491
- Create Singletone to use it as metaclass in ssh_manager.
Singleton provides us posibility of having only one instance of class.
- Create SSHManagaer, wich will manage SSH connections by itself.
Almost SSHClient methods was duplicated in SSHManager.
Now it used so:
SSHManager().method(node_ip, *parameters)
We will not get back SSHClient, we will just say 'Do IT on NODE with this PARAM'
e.g SSHManager().execute_on_remote('127.0.0.1', 'bash_command')
- Use ssh_manager in cli_cluster_deletion test
Change-Id: I307d7d71e814b67d20cc0b4648cf6a7dac4a7829
Closes-Bug: #1503210
The ordering is meaningfull and can mislead in log investigation.
Simple counter added.
Change-Id: I3c2ace4bc0f6d0722a40a0a32086c83f96b7494d
Closes-bug: #1515610
Change default values for networking accordingly
to the existing networks
Apply defaults for new created cluster instead of
reconfiguring all networks
Change-Id: I3eb3c6bdd9128c4e35741884daf0793714ce3b7c
Closes-Bug: 1476185
Co-Authored-By: Vladimir Khlyunev <vkhlyunev@mirantis.com>
Rework fuleweb_test framworks for automation generating TestCase
put yaml confs into fuelweb_test/system_tests_templates
define Case in fuelweb_test/actions_tests
proboscis generate group of test for each case and config
For sen new test plans use Jenkins script.
for all matrix:
sh -x "utils/jenkins/system_tests.sh" -L -j $JOB_NAME \
-t test -w "$WORKSPACE" -i $ISO_PATH \
-o --group=system_test \
-o --show-plan -k -K
for some case and all configuration:
sh -x "utils/jenkins/system_tests.sh" -L -j $JOB_NAME \
-t test -w "$WORKSPACE" -i $ISO_PATH \
-o --group=system_test.create_deploy_ostf \
-o --show-plan -k -K
for some case and some config
sh -x "utils/jenkins/system_tests.sh" -L -j $JOB_NAME \
-t test -w "$WORKSPACE" -i $ISO_PATH \
-o --group=system_test.create_deploy_ostf.1ctrl_1comp_1cinder_neutronVLAN \
-o --show-plan -k -K
for run all destroy controllers failover cases
sh -x "utils/jenkins/system_tests.sh" -L -j $JOB_NAME \
-t test -w "$WORKSPACE" -i $ISO_PATH \
-o --group=system_test.failover.destroy_controllers.3ctrl_2comp_1cndr_neutronVLAN \
-o --show-plan -k -K
Change-Id: I81eeb0df84a0f373be029f20ec6c53e8d3596596
Implements blueprint template-based-testcases
The following CRs provide new mechanism for customizations
setup before master node bootstrap:
https://review.openstack.org/#/c/214749/https://review.openstack.org/#/c/225132/
This patch adjusts system tests to use 'wait_for_external_config'
kernel parameter for customization of master node setup.
Change-Id: Ie89f544b9747255f7a4e75d42367a629fada8dfd
Related-bug: #1485626
Cover case with adding nodes to operational environment
which uses networking template. It should be possible
to update network template after deployment in order to
specify nics mapping for new nodes.
Related blueprint templates-for-networking
Related-bug: #1493391
Change-Id: Ib4d13b31b4b92a30d2dc86ae78c2447522edf415
export REPLACE_DEFAULT_REPOS_ONLY_ONCE=true
- Along with REPLACE_DEFAULT_REPOS=true , enables replacing default
repositories only when Fuel admin node has been installed
export REPLACE_DEFAULT_REPOS_ONLY_ONCE=false
- Along with REPLACE_DEFAULT_REPOS=true , enables replacing default
repositories only before a new cluster is created by tests.
export REPLACE_DEFAULT_REPOS=false
- Disables replacing repositories, Nailgun keeps it's defaults.
Change-Id: Ief45650580cdd897bb785f286d850d27baf06adc
Closes-Bug:#1484137
- Add testcase 'deploy_multiple_services_local_mirror', where
added more services to check more package dependencies during
deploy;
- Move replacing of default repositories from 'empty' to 'ready'
snapshot, to have a clear Fuel admin node with defaults in 'empty'
- Add a new test group 'prepare_release' that can be used instead
of 'setup' to get Fuel admin node with customized settings;
- Remove excess calls for replace repositories from create_cluster()
Change-Id: Iace5351df4f5cb1a3e0c56dade96fd1a2dd7de24
Closes-Bug:#1463793
Bootstrapping 9 slaves at one time reqires a lot of CPU power
(>=10 real CPU cores), so it is better to bootstrap nodes in
two stages to avoid timeouts and libvirt network failures.
Change-Id: I7b3a3ceba4859b6ce4a3af95dbbeba040abd8fff
Closes-Bug:#1455539
- add a new environment variable:
ALWAYS_CREATE_DIAGNOSTIC_SNAPSHOT, default = 'true'.
- move code that resume environment into a separate method.
- rename @log_snapshot_on_error to @log_snapshot_after_test
Change-Id: I5902899ba4c8557b7233e9bf576debc2197ca9e2
Closes-Bug:#1452672
- New options:
EXTRA_RPM_REPOS='name1,url2|name2,url2|...'
EXTRA_RPM_REPOS_PRIORITY=20
REPLACE_DEFAULT_REPOS='true' * Set all provided repositories from
* MIRROR_UBUNTU, EXTRA_DEB_REPOS and EXTRA_RPM_REPOS, as default.
- New methods:
fuel_web.add_local_ubuntu_mirror(cluster_id)
fuel_web.add_local_centos_mirror(cluster_id)
* Add a local directory (inside /var/www/nailgun) as an additional
repository for Ubuntu or Centos.
Change-Id: I57531c835f35cb3d788e6bee6ad9d4129ac16004
Andreas Jaeger