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

This includes:
76972e2 Support a new qpid topology
284b13a Raise timeout in fake RPC if no consumers found
9721129 exception: remove
7b0cb37 Don't eat callback exceptions
69abf38 requeue instead of reject
28395d9 Fixes files with wrong bitmode
bec54ac Fix case error in qpid exchange type "direct"
61c4cde Ensure context type is handled when using to_dict
223f9e1 Clarify precedence of secret_key_file
a035f95 Don't shadow cfg import in securemessage
0f88575 Remove redundant global keyword in securemessage
848c4d5 Some nitpicky securemessage cleanups
5c71c25 Allow non-use of cfg.CONF in securemessage
9157286 RPC: Add MessageSecurity implementation
2031e60 Refactors boolean returns
a047a35 Make ZeroMQ based RPC consumer threads more robust
34a6842 On reconnecting a FanoutConsumer, don't grow the topic name
f52446c Add serializer param to RPC service
5ff534d Add config for amqp durable/auto_delete queues
7bfd443 Avoid shadowing Exception 'message' attribute
Closes-bug: #1178375
Change-Id: Ib5d4733743041eb2324020f9b1dc553260e79b21

Make sure that when spawning multiple instance, the quota exceeded
error still makes sense. Instead of reporting min_count (1) as the
requested number, report "between min_count and max_count".
Fixes bug 1211347
Change-Id: I5a562253788f93b932ccd99bd1dc6630444bd430

This patch changes the behavior of Virtual Power driver. VPD will
now raise an exception should the mac address for the node not
be found. It will also log the node name and mac address.
fixes bug 1201964
Change-Id: I04a275cf8d2cc7665720329000fbe087aa636b21

Create the config drive iso, upload it to the correct location,
then create a VirtualMachineConfigSpec with appropriate contents
and run the ReconfigVM_Task to enable the iso.
DocImpact
Fixes LP# 1206584
Fixes BP vmware-config-drive-support
Change-Id: I1df1e61edbff9f3a68940d3e3e87fea3be838d11

There is a race condition where change_instance_metdata requests still
get to the compute nodes when the VM has a task in operation that means
the VM is no longer present. This change ensures that we skip the
metadata updates when this occurs.
Currently, if this happens, the long running task, like rebuild or
resize will fail because the VM task_state=None and is put into ERROR.
After this change, it will silently fail, and the updated metadata will
be written when the VM gets re-created as part of the long running
operation.
Fixes bug 1207238
Change-Id: I75e1f93e34d3b3ab93a8e8104fd64224f72d7309

Now that the host_ip, for affinity_filters, is coming from the DB and
the host_ip is virt driver independent so make the resource_tracker add
it.
part of bp no-compute-fanout-to-scheduler
Change-Id: I498578191b0af97cb91ca0d81b7ad1eeaf8fd995

Now that nothing is using publish_service_capabilities it can be
disabled, but not removed in order to not break compute rpcapi backwards
compatibility.
Part of bp no-compute-fanout-to-scheduler.
Change-Id: I80c49c46138fd6ee89cb08ffbbced72ada4de72e

Make image_props_filter directly use the data from host_state and not
the data from capabilities that is currently broadcast from the
scheduler over RPC, as this will be removed since it doesn't scale.
Once all capabilities usage in scheduler filters is
removed the capabilities RPC broadcast can be removed.
Part of bp no-compute-fanout-to-scheduler.
Change-Id: I5e6bcec517edbf606903c3e98f3c63f7ba370bcf

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.
This is a second attempt at this patch the first one broke the ability
to filter an arbitrary key value pairs, this time that is supported via
host_state.stats, along with tests to confirm
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: I0b9b05a25a8c1841093913c4a2fb19b48aff2ae1

Some operations require Files-based SRs (ext3/NFS) - such as resize
and VHD download. However, if using 3-part images or RAW then most
operations can be successful with this simple fix.
Fixes bug 1162382
Change-Id: Ia3d1c64871d78ab1bf6e197edaeee170d2b5dea8

When running in dom0, the calls to parted fail to re-read the partition
table so we have to use kpartx to read it specifically.
The failure to re-read the partition table also means that the exit code
is non-0.
Implements blueprint xenserver-core
Change-Id: I56ca25874d6491b0ed70e5efd477254bfe47294d

The `get_pxe_config_file` method is not part of the PXE class.
Change-Id: Ife4f5332af05fd82a865ecdbddf7c5431eb29c1d
Closes-Bug: #1218779 

This patch makes the start_lpar method timeout like the stop_lpar
method.
Change-Id: I111ed7ee80f7a9e786d8cb8b02747b5187120858

If customer deploy a VM with force_hosts or force_nodes options,
then there is no need to retry even if retry filter was enabled.
Fix bug 1211597
Change-Id: I2948f48b5c4982cc29899b76889d7d8f0db47449

To avoid the error in python2.6:
"UnicodeEncodeError: 'ascii' codec can't encode characters",
use safe_encode() to encode basestring instead of str().
Fix bug #1218719
Change-Id: Id26f23ad7f975a3a50750927adbedb86bf80871d

The decorator @classmethod isn't always correctly used in the code base,
this patches fixes some places where it was not used correctly.
Change-Id: Ifd4113b693889de0becc37854f4c62faf2fdab11

V3 tests should using V3 url. Also to make code style consistence, using
V3 HTTPRequest.
part of blueprint v3-api-unittests
Change-Id: Ief1df0bb591941cfaf9555e97bccf21fe1e56594

This patch adds a check in the neutron API for a port create failure due
to over-quota in neutron and raises the new exception PortLimitExceeded.
Also moves the port-create block of code into it's own method to
try and clean up some of the large allocate_for_instance method.
Closes-Bug: #1207914
Related-Bug: #1209446
Change-Id: I4000c8ab550e032363f138a86e1b87f6ab2f5ff2

Just cleans up a small typo in a warning log message that the resource
tracker gives when 'node' is already set on the instance during a claim
operation.
Change-Id: Ic1ece4451f54f4efc8028f8d90de5d240910b20e

The mock vmware objects in fake.py incorrectly represent the
corresponding vSphere server objects. This patch aims to fill
those gaps in the testing framework.
Fixes: bug #1200482
Change-Id: Idc1f93af466e1521c6df655d2851cd923507cc07

We pass the same args to this decorator everywhere, except we have a
different publisher ID for compute api versus compute manager. Use
functools.partial() to avoid this repitition.
Also, this decorator changes with the port to oslo.messaging and this
change reduces the number of places affected by that.
blueprint: oslo-messaging
Change-Id: I54fee992a9995ad3085a6ccd1627958f6dd9ad60

In a bunch of tests, we stub out rpc.call() and use it as a hook to
create instances and return their details.
The thing is, we never actually call() 'run_instance anymore .. we
always cast() it. So, these fake methods never actually get called and
can be safely removed.
A similar change was made in I63bbc98c285faebec53f8e62857c01548807db68
but some more copies of this have been added recently. Copy and pasted
code FTW!
blueprint: oslo-messaging
Change-Id: I1a5a048ed9b0d939d50f7728a348bb6c5a6074dc

Rather than stubbing out catch_client_exception() to bypass the wrapping
of exceptions in ClientException, use the existing ExceptionHelper to do
it.
This is important for oslo.messaging, because there's no public API
method for us to stub out.
This is a continuation of I32c83ef15fece35526739c4fcdfccf7a275c7033 as
another stub_out_client_exceptions() usage got merged by
I4120156db1499dfd3ed22095e528787eb73d33a6.
blueprint: oslo-messaging
Change-Id: If4ff9dba58e9f94fab80be804a98c76fff3267b7

Docker brings an alternative to support Linux Containers (LXC)
as instances. The code has been imported from:
https://github.com/dotcloud/openstack-docker
blueprint new-hypervisor-docker
Change-Id: I0f1ff8acb5a93ba1dc680c68633c43454d1e804c

New SecurityGroupRuleExists exception with 'InvalidPermission.Duplicate'
ec2_code attribute replaced useless EC2APIError.
Note that this error isn't about duplicate group as the code suggests,
but about duplicate permission within a group.
Implements: blueprint ec2-error-codes
Change-Id: I42c94e15a6ca1bc57b10fb683bec17b73de0e0e5