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

Before updating security group rules, we need to make sure that
the info cache is up-to-date. Without this source groups are not
updated properly. This was a regression introduced in commit
85aac04704 which fixed a potential
DOS using source groups.
Fixes bug 1216720
Change-Id: I6b5115df53f2e159ea506ef966cd49cedd35f83d

The following commit(I5b37fa0bb19683fe1642fd81222547d4a317054e) have
not been merged to v3 tree now, because the commit was merged to the
existing api after porting flavors.py to v3 tree.
 commit b65d506a5f
 Author: Russell Bryant <rbryant@redhat.com>
 Date: Thu Jun 27 21:00:05 2013 +0000
 Make flavors is_public option actually work
This patch ports the commit to v3 tree.
Partially implements blueprint nova-v3-api
Change-Id: I3c7f11c53594118456d0cd62eb1ddf74634b77c3

This file was not properly git-add'd to commit:
 f2252de974
Related to blueprint compute-api-objects
Change-Id: Ibd25f4479d49f456bcb49e3eb003edf3d4b2b334

Each object with a list duplicated the _make_list() method in its
own module. This removes that duplication and adds a generalized
helper in objects/base.py. The instance object still uses its own
because it has to do a bunch of other stuff in the loop for
efficiency.
Change-Id: Ic910f39087ebc167b2b930979f7951116caf8598

PCI passthrough device's xml config formatter. adding pci
passthrough device to VM, and detach it from a VM. adapter the
hypervisor PCI passthrough devices management interface given a
unify interface to other PCI passthrough components.
bp:pci-passthrough-libvirt
Co-authored-by: Boris Pavlovic <boris@pavlovic.me>
Change-Id: If55648e5cc2ed5b7ea44aad08205689d99338856
Signed-off-by: Yongli He <yongli.he@intel.com>

Add a parameter to instance_update_and_get_original() to specify the
expecected columns to be returned.
Change-Id: Ib324fdf02f3d462970cdf7c7065ddc3d87f72fba
Signed-off-by: Yongli He

_check_vdi_size in vm_utils.py does not account for VHD overhead. Builds
will fail if on disk VHD file usage is even slightly greater than the
root_gb partition allocation size to the VM.
With VHDs, VM root disks usually come from two-file chains (one, if a
snapshot has never been taken or it was not a fast copy build). There is
a mild file format overhead involved and additionally, once a block has
been written to, the space is never freed or compressed from the VHD
file. The two VHD files in the chain may have overlapping blocks adding
to the total amount of overhead over the actual utilization. This can
result in a VM's partition internally not even being full, with the
external VHD File chain being even just a few MB larger than the allowed
size. Building from an image of such a VM will fail due to this check
expecting external/on-disk usage to never be greater than the internal
partion size granted.
This patch adds a 10GB tolerance which should fix almost all cases of
build failures due to this bug.
Fixes: bug #1216042
Change-Id: I69f47351fe83e54775e0147a27450c21196f9bfd

Blueprint: hyper-v-ephemeral-storage
This patchset adds support for creating an ephemeral storage dynamic
VHD or VHDX (based on the versions of vhdutils used) when an instance is
spawned.
The ephemeral disk is correctly migrated during migration/resize and
live migration.
The ephemeral virtual disk is not partitioned or formatted, leaving the
task to guest side tools (e.g.: cloud-init or cloudbase-init).
Change-Id: I0b5d6b55ccdc981efcd507c45184fb2d528bf9c1

Blueprint: hyper-v-vhdx
The VHDX virtual disk format has been introduced with Hyper-V Server 2012
providing performance and management improvements.
Unlike the VHD format, VHDX allows to resize differencing disks. This
reduces the requirements for local disk space in case of CoW images and
improves the performance of the first spawn operation for any given image
and flavor combination.
Supporting the VHDX format in Hyper-V requires the WMI V2 namespace.
Unit tests have been added accordingly.
Change-Id: Idddec06b9639e22d69426562d1b5a0cbf2890ccf

Creates a new DBObjectMixin class for DB-backed objects to use. This
allows us to create non-DB-backed objects without having the common DB
fields automagically added to them.
Related to blueprint unified-object-model
Change-Id: Iaf2d8500505e9acdbffffb1d4bd3db0870ae82a3

It's no longer used. Deprecate it.
Related to blueprint unified-object-model
Change-Id: Ib7be13c53231220eec232442cb1836626b4b580f

Converts calls to compute's finish_revert_resize to pass new-world instance
and migration objects.
Related to blueprint unified-object-model
Change-Id: Iee4a47b3dbe96f66c061b51ae9512dad9b86a870

Converts calls to compute's finish_resize to pass new-world instance
and migration objects.
Related to blueprint unified-object-model
Change-Id: Ic513b826087ced39c8aa939e93fe6c0d250dc1a5

Converts calls to compute's resize_instance to pass new-world instance
and migration objects.
Related to blueprint unified-object-model
Change-Id: Idad237ca76b4a28dc80792ad195b95db224161ae

Part of blueprint vmware-nova-cinder-support
This patch adds support for booting from a cinder volume.
Change-Id: Ic2104ddda759dbfb3f352b547b417c910d75d9db

To allow a single VC driver to model multiple clusters in vCenter
as multiple nova-compute nodes. The VC driver will be configured
to represent a set of clusters as compute nodes.
For example to specify two clusters named cluster-A and cluster-B,
the nova.conf should have:
 cluster_name = cluster-A
 cluster_name = cluster-B
DocImpact
Change-Id: Ia5464948cc30b9b744450f9c301c4f3afaff717b
Blueprint: multiple-clusters-managed-by-one-service

Part of blueprint vmware-nova-cinder-support
This is the first part of the blueprint that deals with attachment
and detachment of volumes.
Change-Id: I3ad8ae127eef7bfe600be91629c9352a7a5fa4b4

Blueprint: hyper-v-dynamic-memory
Memory overcommit in Hyper-V is handled with a feature called dynamic
memory (memory ballooning). This change adds a configuration option to
enable it.
The value of dynamic_memory_ratio expresses the ratio between the total
RAM assigned to an instance and its startup RAM amount. For example a
ratio of 2.0 for an instance with 1024MB of RAM implies 512MB of RAM
allocated at startup.
Unit tests have been added for this feature as well.
DocImpact
Change-Id: I036eaa3e739d995c6a5bcfb022bf2ed0a999085e

In case an image without, or with an old cloud-init is booted, decide
if it's the file, or the nbd device that needs to be resized. Qcow2
files have to be mounted as nbd before attempting a resize.
Fixes bug 1212734
Change-Id: I29cbd4b334c39936342fa429bcc0b95378cade36

The pattern for the 'sudo nova-rootwrap' command was duplicated
throughout the code. Put it in one place in nova.utils.
Change-Id: I0d4edfa5a53c3b9a9fa43864f5daaa17a0d024a1

When creating an instance, a few properties from the image will be saved
in the instance's system_metadata under the prefix 'image_' for latter
usage.
The keys that the instance will inherit from the image are: 'min_ram',
'min_disk', 'disk_format' and 'container_format' together with the
inheritable image properties.
Related to bug #1039662
Change-Id: I358fb6d8dcafd67c289fe5083bb82140f3759fee

Removes rogue subprocess.Popen calls to make better use of
nova.utils.execute where appropriate to benefit from the
patches pushed into oslo.
Fixes bug 1053382
Change-Id: Iaee5379edd51f215a51d55263196dd6c0d2de0c3

Fix bug #1220454
In the first version of the IOVisor driver, versioning was included in the
the PATH of the native IOVisor commands which was /opt/pg/bin/0/
In the current version, versioning has been deleted from the PATH and
now it should only include /opt/pg/bin/
Change-Id: I2b88eb28593043843a9ac0afd5bff7c268db0f60