* Try /etc/tempest DIR if the config file not found
* Exit immediately if the config file not found
* Remove pika from dependencies
* tempest named starter script (pep8 removed run_tests.sh)
Updates the test_volume_create_get_delete test so that it no
longer tries to cleanup when the volume variable is None.
This resolves UnboundLocalError errors which can occur in
the test suite when volumes aren't getting created correctly.
In some cases the wait on volume deletes will receive a 404
error which results in a log output. However for these tests
this is the expected behavior so the log output isn't needed.
This adds a flag to suppress log output on 404 errors, which
is then used on all wait functions.
Fixes: bug 1084283
Not all configurations of OpenStack have the EC2/S3 API's
enabled. For that matter the S3 "API" in Nova (object-store)
isn't even production ready. As such it would be really nice to
hide exceptions if the S3 and EC2 clients in Tempest can't be
This commit adds a test case that checks if all the absolute limit
values are returned in the REST call to Nova API Extension
"/limits". Also modifies limits_client.py file for JSON and XML
requests so as to make it less verbose. 'get_specific_absolute_limit'
method is added in the limits_client.py file. As method
'get_personality_file_limit' is removed, "test_server_personality.py"
Fixes: bug #1025294
* Adds a client for the 'os-quota-sets' extension
* Adds basic Admin and non-admin tests for GET and PUT operations for the
* Adds some tests to check Quota enforcement for create server (bug 1034453)
Fixes LP Bug #1040760
Fixes LP Bug #1034453
Continues the effort of the https://review.openstack.org/#/c/3064/
* add EC2 keypair and volume tests
* add EC2 snapshot from volume test
* add EC2 floating ip disasscioation
* add EC2 operation on security group
* add EC2/S3 image registration
* add EC2 instance run test
* add Integration test with ssh, console, volume
* add S3 object and bucket tests
Mysql treats 0 and False as the same thing, but this isn't true for
postgresql which is much more strongly typed. Our whitebox testing
was failing on postgresql because of this assumption.
Tempest was catching a generic fail on the metadata too
long call, but a very specific error is supposed to be returned,
a 413. In real runs there was a bug in nova which crashed on the
request (fixed in https://bugs.launchpad.net/nova/+bug/1079387).
Fix tempest to only accept a 413 as valid on this request.
Make this an assertRaise test instead of a negative test
we were always adding 'tempest' to the list for tests in
run_tests.sh, which didn't let you use the wrapper for running
a single tempest test. This change should fix that.
Probably could be cleaner, but my bash-fu is weaksause.
* Renamed services/nova to compute
* Added OpenStack license where missed to compute services/tests
* Broke down compute tests into separate packages by functionality
Updates the Tempest common rest client so that it sets the
'message' variable value in all cases. Previously some exceptions
could slip through and cause 'message' is not defined errors.
Updates tempest/manager.py so that it ignore's ImportError
exceptions when quantumclient is not present.
Not all Tempest tests require a quantumclient so it makes sense not
to require it all the time. If a user tries to run quantum specific
tests they will simply get an error when trying to create the
network client instead should quantumclient not be present.
Related to bug 1051881.
Two test cases: one to cover the live block migration on XenServer, and
the other to indicate that the machine status is left in an inconsistent
state in case a non-existing host is specified as target.
To run the live-migration tests with block migration, include:
live_migration_available = true
use_block_migration_for_live_migration = true
in your tempest.conf
To run these tests, the hypervisors must support the Storage XenMotion
test_list_servers_negative had so many issues with it, it's hard
to list them here... but at least the following has been fixed:
* No longer will the test flatly delete all instances the tenant
and alt tenant have during every test method run (!!)
* Fixtures -- 2 active and 1 deleted server -- are created in the
class' setUpClass() method instead of having test methods
constantly re-launch instances -- this reduces the test time
down about 3 minutes
* Removed all non-negative tests
* Removed all tests that were skipped due to bugs that have been
* Skip the XML create_service() test for Identity Admin tests
because of the upstream Keystone bug 1061738
This patch also includes patches for bug fixes that Sean Gallagher
contributed but needed to be rebased into this one in order
to get the full gate to pass.
fixes: LP bug #1059802
fixes: LP bug #1060373
fixes: LP bug #1061740
fixes: LP bug #1061167