This moves some conductor compute task tests into the test class that
only tests the manager side. It also removes testing of
scheduler_utils's set_vm_state_and_notify method as that is already
tested separately.
Change-Id: I30bd3897311870b10c8ec65182c81d341a904485
They were missing... although there is some testing via some conductor
tests that are not completely unit-y. Those will be adjusted next.
Change-Id: I696dd8ff264fffb3ae015c3ae653c0799c60cecf
The tests for the admin_actions API extension (or v3 plugin) are
stubbing out conductor methods, which creates a lot of pain when
compute_api internals are changed. Convert these tests to unit tests,
checking only that the appropriate compute_api methods are called with
the correct arguments.
Change-Id: I16634efa2397b36afa15b90712bfd678325bc049
Move resize related compute API tests to test_compute_api and make them
unit tests. This makes test changes easier when converting to objects.
This also caught a couple things where the cells compute api class was
not in sync that don't really end up being bugs, but need to match for
tests to pass.
Change-Id: I194310beb5e821e4a8262310d1bb80e0747e1c43
Fix bug 1212648
When boot multiple instances in one request, the launch_index
was not increased.
The fix was increase the launch_index one by one for each instances
in one request.
Change-Id: I27dde8cfae52853d33277c3d8e0e65c91bc6280b
This was deprecated in Grizzly, and the TODO said remove for Havana.
fixed_range was deprecated in I4ec111079f7a1d253190e6a6008048f992a53f68
DocImpact: Remove deprecated config option
Change-Id: I1f429de7a7091adbd3a178d287b3a90e02664450
Now that compute/api.py:update() is no longer being called for instance
updates the policy check is no longer enforced. This adds policy
checking back in at the API level so that the compute:update policy will
continue to function as it did before.
Bug 1212807
Change-Id: I365ad8ac2845d914deefb841c314a50f4869dba0
There is no implementation of _action_change_password for the
attribute of ActionDeserializer in V3 server API. Get rid of it
fix bug 1209072
Change-Id: I37019811caa71c491c228b99babfcfb34bbaa685
Looks like the lack of a test case sneaked in this bug. Added a
simple one that just ensures that we can get a list of all the
metadata information. Basic issue was that the check_version
was checking against ec2 versions. Changed it to check OpenStack
version including the 'latest'.
Fixes bug #1211122
Change-Id: Ifac7659467b81abcffd3f84a24292c11f2e058fc
Make compute_capabilities_filter directly use the data in host_state and
not the data capabilities data that is currently broadcast from the
schedulers over RPC, as this will be removed since it doesn't scale.
The filter name, compute_capabilities_filter, is kept the same to keep
backwards compatibility.
Once all the capabilities usage in scheduler filters is removed, the
capabilities RPC broadcast can be removed.
Part of bp no-compute-fanout-to-scheduler.
Change-Id: I846259b363a794803e230eaf25c69c52970b6c4f
Instead of using host_ip from capabilities, which are being removed, get the
host_ip from the same place the rest of scheduling information comes
from, the database.
Part of bp no-compute-fanout-to-scheduler
Change-Id: I56a1cbb1903c20b347469d0ee5a8c6d89a21c44e
In order to remove compute capabilities being broadcast via RPC to the
schedulers, host_ip and supported_instances, the two pieces of data the
scheduler uses that aren't in the database today, need to be added.
In the scope of this blueprint, only the way the filters are getting
their data is changed, not how the data is generated or anything else.
host_ip is used by the affinity_filter and supported_instances is used
by image_props_filter.
Part of bp no-compute-fanout-to-scheduler
Change-Id: I72371260159c9fdffcdb57209095fc103b6dd79e
Add 'supported_instances' to the function get_available_resource for all
virt drivers (except libvirt which is done separately). This will help
make image_props_filter work without the RPC broadcast capabilities.
Part of bp no-compute-fanout-to-scheduler
Change-Id: If9eee203b3cbb12963dc55e4215d206cb5d81264
get_available_resources is actually used by the scheduler while
get_host_stats was previously used, and is still used for host
capabilities. This patch makes get_available_resources and
get_host_stats use the same logic to clean up the code. As part of
making them use the same logic, some of the unused data returned from
get_host_stats is changed to be what get_available_resources expects.
This is also in preparation for removing the periodic RPC fanout from
compute nodes to the scheduler.
This patch cleans up libvirt and fake drivers only. This is not needed
for other virt backends as this is a cleanup only. Further cleanup of
libvirt and other drivers will happen after get_host_stats isn't used
for the compute fanout to schedulers.
Because this makes a change to the fake driver, several api samples
needed to be changed as well. The fake driver is changed so the
test_virt_driver tests can continue to be used
Part of bp no-compute-fanout-to-scheduler
Change-Id: I1eec5c117a1cb0490e9f9c09e731909bc31698a9
There were several methods overridden from the base virt driver class in
the powervm driver but all they did was pass which can cause problems
with tempest, e.g. the tempest test for pausing an instance will timeout
waiting for the instance state to change. This change raises
NotImplementedError rather than passes so that tempest will fail out
quickly when testing something that the powervm driver does not
implement.
Also removes the overridden manage_image_cache method since it didn't do
anything, which is exactly what the base class does.
Added unit tests to make sure that methods which should simply pass do
and that tests which should raise NotImplementedError do that also.
Fixes bug 1205066
Change-Id: Ieaf256a5c0f54804686318d6cbd5877a5003c9eb
If gluster is present in qemu_allowed_storage_drivers, glusterfs's
backend will pass a disk configuration to QEMU. This allows QEMU to
access the volume using libgfapi rather than mounting GlusterFS via
fuse.
Implements blueprint glusterfs-native-support
DocImpact
Co-authored-by: Flavio Percoco <flavio@redhat.com>
Change-Id: I939cac05ee333b77abcaa9bd132d5eb3308e60ff
lawrancejing