openstack/nova-specs - nova-specs - OpenDev: Free Software Needs Free Tools

This will provide a mechanism for the user to tag a device they
have assigned to their guest with a specific role. The tag will
be matched to the hardware address of the device and this mapping
exposed to the guest OS via metadata service/cloud-init
Spec for bp/virt-device-tagged-attach-detach
APIImpact
Previously-approved: Newton
Change-Id: Idcac1506b61c29f02e335200fb18d394243098e2

This was a feature from Nova legacy API codebase that had 0 testing,
and only hints of documentation. This proposes to bring it back in a
very limited (and deprecated way) for the current code.
Change-Id: I70a8c8e76f234611f4b2972b3adae86ca09da012

This removes the portgroups functionality, as we didn't/won't get to
that this cycle. This will be proposed as a separate blueprint in a
future cycle.
Change-Id: Ic97c172c72245ffa9fe8c2544639a1cb7a898e4d

Modify 'filtering/cliaming' to 'filtering/claiming'
Change-Id: Idec6297e62543a90cf3e0b5a3730544cbd644389
Closes-Bug: #1610092 

Populates the allocations table in the API database with usage
information by calling the new placement API from the Nova resource
tracker.
Change-Id: Ia90ddb7d148f5ec9cb67cf378b93784371af82c7
blueprint: resource-providers-allocations

This re-writes the ironic-multiple-compute-hosts spec to use a hash
ring, rather than messing around with how we schedule.
Change-Id: I51de94e3fbe301aeed35a6456ed0b7350aefa317

Not sure how or why we forgot to include an ability for a caller to set
and delete allocation records against resource providers, but we did.
This amendment to the spec adds two additional REST API calls for
setting and deleting allocation records in an atomic fashion (atomic
operation for a set of resources consumed by a single consumer against
one or more resource providers). These REST API calls will be made from
the Nova resource tracker when instances claim or free resources on the
compute nodes.
The endpoint is /allocations/{consumer_uuid} accepting PUT and DELETE
methods.
Co-Authored-By: Chris Dent <cdent@anticdent.org>
Change-Id: I5ef70ae2e13b1dd0c0ef1fa5821b3ee5a3265ca7
Blueprint: generic-resource-pools

In-development work has shown that an additional method is
needed in the resource providers API to allow the setting
of multiple inventories on one resource provider, in one
request.
This has been added as
 PUT /resource_provideres/{uuid}/inventories
with a body of one or more inventories.
Change-Id: I81f1b26579e8c2b1aa4afce3a90979ea7a925008
Blueprint: generic-resource-pools

A bug was found where passing an invalid status to the server listing
API would return an empty list for a regular user, but would raise an
exception and return a 500 error for admin users. In the discussions
about that bug it was felt that while fixing the exception was certainly
correct, the list of valid status values is small and well-defined, so
returning an empty list was not the desired behavior; instead, the user
should get a 400 Bad Request instead.
Blueprint response-for-invalid-status
APIImpact
Change-Id: Id5a9b6708b39530ca6f53e9c576305d01f211c78

This is rendering poorly, fix it up so I can read it. :)
Change-Id: I3584c3067d3f8da9be821cf10579ca91959e2d61

This tries to put down the complete arc of API extensions use in Nova,
and how we will be tearing down the infrastructure over the next 2
cycles.
Change-Id: Ic44397fbaf142816091eb1817dcbea249b3528b0

Amends the generic-resource-pools spec with the following changes:
* Change terminology from "resource pool" to "resource provider"
* Change the REST API proposal to have /resource_providers endpoints
instead of /resource_pools endpoints
* Change the `openstack resource-pool` CLI commands to `openstack
resource-provider` CLI commands
* Adds a generation field to the output of the resource provider dicts
and update operations along with a note as to how the generation field
shall be used
Change-Id: Iefe28041fa9e81558ebac9f47016c4db43a92723

Currently, no useful notification will be sent
for keypair state change.
This spec will transform legacy notification
to versioned notification about keypairs,
and at the same time extend contents of
notification with extra data.
Blueprint: keypair-notification
Change-Id: I05080afb8bd4249f8bea23512b25844fecbefeb9

The spec is describing the Nova side of the 'multiattach' feature.
The functionality targets to add support for attaching a Cinder
volume to multiple VM instances.
Previously-approved: mitaka
Implements: blueprint multi-attach-volume
Change-Id: I288f7140e3acd979b65734049fd42c2b3f3c5277

The deadline for non-priority specs for Newton is almost here,
and as always there will be a large number of specs that do
not get approved. Any that remain in the Newton directory will
eventually get force abandoned. Opening up the Ocata directory
allows submitters who wish to re-target their effort to the
subsequent release, to move their spec across to the new
directory straight away enabling ongoing visibility & review.
Change-Id: I732576bb3c0e60ec130cdf73647fbc2f234905f7
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>

In Neutron, there is a effort to support routed provider networks. A
routed network in this context, is a physical network infrastructure
that implements scaled networks by routing instead of large L2
broadcast domains. For example, deployers may have routers at each
top-of-rack. Instead of a single VLAN covering the deployment, each
rack would have its own VLAN and the routers will provide reachability
to the rest of the racks over L3. Operators want Neutron to model
this like a single network. This has implications for Nova
scheduling.
Depends-On: I1a71e83d9fb3d13306979934fbd11faf80091ab8
Change-Id: I8f5523930d360a9183ab111e015ef4b052e0968f
Related-Bug: #1458890 

This blueprint is to add nova an ability to support perf event, these
event data can be collected by Ceilometer.
Co-authored-by: Alex Xu <hejie.xu@intel.com>
Blueprint support-perf-event
Change-Id: I7cec5c583048134b1a7c3986428e32672434c3c8

https://blueprints.launchpad.net/nova/+spec/discoverable-policy-cli
Not all users have the same set of privileges. Not all API endpoints
are enabled. Because of this, adding an nova-manage CLI subcommand that
lists the operations an user is allowed to use can be helpful. This spec
describes phase one of a two phased implementation. The first phase
will prove out the idea using a raw representation that will be
improved in the second phase.
Co-Authored-By: Chris Dent <cdent@anticdent.org>
Implements: blueprint discoverable-policy-cli
Change-Id: I973f1e0545c73d66ef17bed77c6dda5de007cb1e

If the *serial console* feature is enabled on a compute node with
``[serial_console].enabled = True`` it deactivates the logging of the
boot messages. From a REST API perspective, this means that the two
APIs ``os-getConsoleOutput`` and ``os-getSerialConsole`` are mutually
exclusive. Both APIs can be valuable for cloud operators in the case
when something goes wrong during the launch of an instance. This
blueprint wants to lift the XOR relationship between those two REST
APIs.
blueprint libvirt-virtlogd
Change-Id: I9a1fbf005b0f48df90093a346579d9ddc64f7846

For some reason a number of specs (and docs) have zwsp characters
inserted in URLs. This causes crud in the reviews, and really
confusing errors around line lengths for people (we count them, but
they aren't seen in their editor).
This removes them and makes a test to make sure they never come back.
Change-Id: Ib6e94725e2716dbb2ff86f272fbe4a6987714444

blueprint cells-quota-api-db
Change-Id: I7253809253911d64e8637e7b1731889702bcfe3e

Add new optional parameter offset for console-logs
command.
APIImpact
Change-Id: I1783ce64619196af7741e46e7abfd8afbe1a3b67

live_migration_compute_communcation.rst formatting
is fixed. Minor rewording for the paragraph was done.
Change-Id: I51dffbdc45c337f324b1b57c67895de819f80a01

When an instance is rescued the libvirt driver creates a file in the.
instance directory that contains the original instance xml. This file
is used when unrescuing the instance to revert to the original disk
configuration. However, this file is not necessary to restore the
instance's original configuration. Furthermore the reliance on this
file means we currently cannot live migrate a rescued instance. Thus
it is proposed that we do not create this file during the rescue
operation and rebuild the instance's original xml file from the
database when we unrescue it.
APIImpact
Change-Id: Id3a055f00c3f8ef006911e2523c8d284f449b524

Spec to automate the use of various libvirt/qemu
features to complete live migrations when instance
memory access rate is high.
Change-Id: Id0b334ba8c4d97b3eaa58a6fe3b1242560567853
Blueprint: auto-live-migration-completion