skip test_lock_unlock_server due to bug 1233026.
Fix the bug will change the exception of self.servers_client.stop.
After this bug fixed, will update this test.
Change-Id: Ie1e8de6b3582147a44337b0cd6f0ab51531b6938
tempest/config.py has many simliar tiny functions for register_*_opts.
We should remove the code duplication to providing a common function.
Change-Id: Ie8facc21a2d8a1e27687b1182a3db9f98bc1a05e
Closes-Bug: #1236498
Sometimes the nightly stress test run fails. This will catch all
exceptions in run_stress.py and log them.
Change-Id: Ib6abd2c157df3988ccae1c6c8f703ec0ed24ff00
Fixes-bug: 1234580
Change I8519cba1fcf6ff39d69839783d07359544509db5 was supposed to make
the setup for test_list_servers_negative.py handle the case that tempest
is being ran without tenant isolation, but it doesn't fix the entire
problem because it's only checking if the MULTI_USER global is False.
The MULTI_USER global is only False if allow_tenant_isolation=False and
user1 == user2. So if you're running tempest without tenant isolation
but you do have two separate users/tenants for running the tests, then
the setup can still fail.
This patch fixes the problem by always doing the cleanup if
allow_tenant_isolation=False regardless of what MULTI_USER is set to.
Closes-Bug: #1234370
Change-Id: I821e200d07c11ba04d8be788095b7648de1d820b
If we use skip_because decorator, the below error occurs.
_StringException: ImportError: Failed to import test module:...
And, all tests of the class having the skip_because decorator are not
tested.
This commit fixes it.
Change-Id: I98e51d3bfd8b13dc3ef462eb92b170295f5f142e
Closes-Bug: #1236220
The setup method of test_volumes_actions generates a random name for the
instance and volume it uses for the tests but the names don't include
the test class so if they don't get cleaned up it's hard to debug where
they came from. This patch simply adds the class name to the
instance/volume name generation.
Closes-Bug: #1235411
Change-Id: If457e73c11f08fd57ee77a85208de84fc39eda8c
The test_list_server_filters test class was doing a lot of
setup/teardown work that is now handled in the parent class. We still
have to explicitly set the name since some tests use it for filtering
but otherwise we can rely on the parent class to do most of the heavy
lifting here.
This change makes it so that the instances have a name based on the test
class rather than just a random integer which is hard to use for
debugging leaked instances.
This change also removes the get call from setup and moves it into the
test cases that actually need to use it (which are currently being
skipped anyway).
Closes-Bug: 1235411
Change-Id: I66813c900059dd4232de645c225032e58e9aecde
When unseting a nonexistent key, an exception should be raised.
when settting flavor extra specs, key is arbitrary, so
'set_nonexistent_key' does not make sense
Change-Id: I1b94a8b9524ed67dfc3fb1afd51442b740873638
this change introduces a skip_because decorator which accepts two args
a bug and a condition; also updates the skip tracker accordingly
Change-Id: If53f2ef81d6bddbce284267216254b467046855f
The test_list_servers_negative tests require that there are no existing
instances in the database when they run and will fail if any are found
under the primary/alternate user/tenant.
When running tempest with allow_tenant_isolation=false on a slower
system, there is a race where instances from previous tests are still
cleaning up in the background and cause test_list_servers_negative to
fail in setup.
This patch makes test_list_servers_negative block while the deleted
instances are being cleaned up in the background if running without
tenant isolation.
Closes-Bug: #1234370
Change-Id: I8519cba1fcf6ff39d69839783d07359544509db5
This change includes the following improvements to the
autoscaling test:
- On stack create status timeout, override the error status so
that FAILED stacks fail fast
- Improve the error message in assertScale
- Delete the stack at the end of the scenario, and confirm the delete
succeeded
Change-Id: Ia220e3a2b422472ac7c6d29f9a5a61bd7c4987c3
The error status and not-found exception used for
status_timeout and delete_timeout are specific to nova.
This change makes it possible to override the nova defaults.
Specifically, this will improve the status polling of the
orchestration scenario tests.
Change-Id: Ib55f95ed51ecce3813a57c330e0f936414ec3ff2
The test_list_servers_by_changes_since test fails when running with
allow_tenant_isolation=false because there is an arbitrary 2 minutes to
determine the timedelta for the changes-since in the query. That doesn't
work when the servers created in the test are all under the same
user/tenant because the tearDown doesn't block until the servers are
deleted.
This patch makes the timedelta code smarter in that it narrows the
changes-since down to when the test case started setting up the servers
for the test and when it does the query.
Closes-Bug: #1195309
Change-Id: Ie848637ea490da2151e5db8292726111d45a8c2d
Creating xml version for the neutron router and floatingip related
operation. The test_floating_ips.py also extended with the XML variant.
Change-Id: I5e85cd263828a127ed48db04b19fbf4d145f1d87
add an expected behaviour test of deleting server
- delete_active_server
add two negative tests for server unpausing
- unpause_non_existent_server
- unpause_server_invalid_state
Change-Id: Ia9b5fdf1c2298ff685322fc99a5496710a7a67ef
