This mirrors the indirection fixture override that we have for
object_class_action() while we transition fully to the base
VersionedObjectRegistry. This prevents us from breaking with current
oslo.versionedobjects.
This also caused a few other tests to get run with manifest calls,
which needed to be converted to continue to work.
Change-Id: I2aaa42784b2cc5c6898a3fac0c951dd57dd4f4c6
Closes-Bug: #1516805
The new test_schedule_to_all_nodes test fails in some epic ways in
cells. Not really worth debugging, this is a new test and had just
exposed another cells / devstack issue. Just skip the test.
Change-Id: Icc71e36f4ecb015dff9e806caacd31262f7e17f7
The original support for Opaque networks make use of a
configuration flag indicating that a global integration bridge
was configured on the ESX host. If this global integration
bridge was used then that would be selected as the opaque network.
The next generation NSX neutron plugin will not make use of the
integration bridge. This will require that we make use of the network
ID for the opaque network.
The support for VC 6.0 has specific support for this. For 5.5 we make
use of a DynamicProperty setting to achieve the same goal.
DocImpact
For backward compatible opaque support the setting
'CONF.vmware.integration_bridge' needs to be set.
If this is not set then the network id will be used as the
opaque id.
In addition to this the default value has been set to 'None'.
Implements blueprint vmware-expand-opaque-support
Change-Id: I309ca2bc8186b8eefc0a979d039dd4cd2a6f89f6
Right now to check if some part of code supports
requested microversion developer usually:
1. get a microversion from request.
2. create instance of class APIVersionRequest
with requested microversion.
3. compare these two objects.
This check takes at least 2 lines and looks ugly.
Sometimes developers create unnecessary class fields
to store object APIVersionRequest with requested microversion.
To make nova code more readable and simple method
'is_supported' was created.
Microversion check with this method takes only one line
and code looks more beautiful.
Change-Id: I9078cfa1afad8b6b09583dac87fa757a0cc32daa
Add log hint to info message. In addition to this follow the logging
convention of using a , instead of a %.
TrivialFix
Change-Id: Ie9992835420aacf65b0962b51581e0e05c5bac72
Add html for sphinx invocation so that files end in this directory
as usual. This is the usual place for sphinx documentation.
Change-Id: Iae7a3b91d407c06b7f4d2dce909a15c38ea72b9b
If volume detach fails after reconfiguring the VM to remove volume
vmdk then it will not be possible to detach the volume again. This
is because the VMware driver fails the detach operation by raising
StorageError when it finds that the volume disk device is missing.
Due to this, the volume state will remain 'in-use'.
The compute manager expects the driver to raise DiskNotFound if the
disk is missing during detach operation. It ignores the exception
and allows the detach to complete. This patch fixes the problem with
the VMware driver by throwing DiskNotFound instead of StorageError
if the volume disk device is missing.
Change-Id: I4074d240a3468534c0a269f6cd160f399e8fa18c
Closes-Bug: #1514910
In the resource tracker the pci stats is show in
the log like pci_stats=PciDevicePoolList(objects=[PciDevicePool]
the patch fix it to be logged as
pci_stats=[PciDevicePool(count=7,numa_node=None,product_id='1004',
tags={physical_network='physnet1'},vendor_id='15b3')]
Closes-Bug: #1513467
Change-Id: I9434bb5cf0ed9eca291df6f4a8b8568b7c3ff4eb
Jenkins