| Commit message (Collapse) | Author | Age | Files | Lines |
|\
| |
| |
| | |
proposed/juno
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
commit a8a5d44c8aca218f00649232c2b8a46aee59b77e add pci_request_id
as one item for the request_network tuple. But the icehouse code
assume only three items in the tuple.
This patch filters pci_request_id out from the tuple.
Cherry-Pick from:
https://review.openstack.org/#/c/126144/6
Change-Id: I991e1c68324fe92fac647583f3ec8f6aec637913
Closes-Bug: #1377447
|
|\ \ |
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Commands run:-
$ python setup.py extract_messages
$ python setup.py update_catalog --no-fuzzy-matching \
--ignore-obsolete=true
$ source \
../openstack-infra/project-config/jenkins/scripts/common_translation_update.sh
$ setup_loglevel_vars
$ cleanup_po_file nova
Change-Id: I64b2b468f7edd44dbb445b5b4e68b65c3fa53d9e
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Cert revocation was broken by
32b0adb591f80ad2c5c19519b4ffc2b55dbea672. os.chdir() never returns
anything, so this method would always raise an exception. The proper
way to handle an error from os.chdir() is to catch OSError.
There were existing tests for this code, but they conveniently mocked
os.chdir() to return values that are never actually returned. The
tests were fixed to match the real behavior.
Change-Id: I7549bb60a7d43d53d6f81eecea31cbb9720cc8b6
Closes-bug: #1376368
(cherry picked from commit c8538208da00c3b0d0646629c9d668aa69944b85)
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Update all of the rpc client API classes to include a version alias
for the latest version implemented in Juno. This alias is needed when
doing rolling upgrades from Juno to Kilo. With this in place, you can
ensure all services only send messages that both Juno and Kilo will
understand.
Closes-bug: #1378786
Change-Id: Ia81538130bf8530b70b5f55c7a3d565903ff54b4
(cherry picked from commit f98d725103c53e767a1cddb0b7e2c3822309db17)
|
|\ \
| |/
|/| |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
When a ProcessExecutionError is thrown by processutils.ssh_execute(),
the exception may contain information such as password. Upstream
applications that just log the message (as several appear to do)
could inadvertently expose these passwords to a user with read access to
the log files. It is therefore considered prudent to invoke
strutils.mask_password() on the command, stdout and stderr in the
exception. A test case has been added (to oslo-incubator) in order to
ensure that all three are properly masked.
An earlier commit (853d8f9897f8563851441108a9be26b10908c076) failed
to address ssh_execute(). This change set addresses ssh_execute.
OSSA is aware of this change request.
Change-Id: Ie0caf32469126dd9feb44867adf27acb6e383958
Closes-Bug: #1377981
|
|/
|
|
|
|
|
|
|
| |
EC2 describe images crashes on volume backed instance snapshot which has
several volumes.
Change-Id: Ibe278688b118db01c9c3ae1763954adf19c7ee0d
Closes-bug: #1370265
(cherry picked from commit 1dea1cd710d54d4a2a584590e4ccf59dd3a41faa)
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Converting network_get and network_get_all to use objects broke
the display of the os_networks extension, because IPAddress
fields in Network objects are dumped as lists by the jsonutils
extension. We therefore must explicitly convert these object
field values to string.
The tests are updated to use objects so that we pick up bugs
like this in the future. Incorrect assertEqual parameter order
is fixed in the tests too since these are comparing dicts and
it's not fun debugging a MismatchError when the reference/actual
values are backwards.
Change-Id: I0f05a9b4d7bbe5fe0a3b110c191455ca7edefcb5
Closes-Bug: #1376945
Co-authored-by: Matt Riedemann <mriedem@us.ibm.com>
(cherry picked from commit da25467aafce9b62dd3fdff9d6cd84121fbee17e)
|
|\ \
| | |
| | |
| | | |
proposed/juno
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
If you're not at a new enough version of libvirt, the compute service
fails on startup because VirtNUMATopologyCellUsage is not fully
populated.
This add a min version check before trying to get host NUMA topology
information.
Closes-Bug: #1376307
Change-Id: I00f6325cb554bc5e34d9f0fe651af39630f35b5d
(cherry picked from commit 8ba0d9188d492028fcf4e65f908aa2d3db571952)
|
|\ \ |
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Patch I7598afbf0dc3c527471af34224003d28e64daaff introduces a
Minesweeper failure, due to the fact that it doesn't distinguish
between destroy operation triggered by the user and by the revert
resize.
This patch fixes the issue by checking the task state. If the task
state is revert_resize, the original VM doesn't get deleted.
Closes-Bug: #1376492
Change-Id: Idb9ac6c1ec5dcea52ce8e028f5cce08da1779321
(cherry picked from commit e464bc518e8590d59c2741948466777982ca3319)
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The integrated api client previously did the HTTPConnection /
HTTPSConnection url parsing dance. In python 2.x HTTPSConnection
doesn't care about SSL certs at all. While not actually an issue for
these tests, it does mean we keep around an example in the code that
uses HTTPSConnection, which will prevent us from creating a hacking
rule to keep those out once the other 4 actual security issues with
HTTPSConnection are removed.
Change-Id: Idd7d5a055600dda663f9c56b39883510f8688b12
Related-Bug: #1188189
(cherry picked from commit 777a5870c9f29949e6af704bfa03c2e204065ab1)
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
| |
TrustedFilter was using httplib which doesn't check for CAs.
Here the change is using Requests and verifies local CAs by default (or another
one if provided)
This effort is related to CVE 2013-2255.
SecurityImpact
Closes-Bug: #1373993
Change-Id: I0b8e6319a4cc39876b1e396ef705f0fc5def1e44
(cherry picked from commit 30871e8702737edbbfbcbbb5f21858873b37685c)
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This patch makes libvirt report memory in Mb (as
hardware.VirtNUMATopologyUsage class expects Mb) and that when
generating XML we use Kb as this is what libvirt expects by default.
Change-Id: Ic4518acf6bcee463009437829646de8c83aff6bf
Closes-bug: #1373159
|
|\ \
| |/ |
|
| |
| |
| |
| |
| |
| |
| |
| | |
We should pin each vcpu into specified numa's cpuset. But the current
code only pin the last vcpu in one cell to specified numa's cpuset.
Change-Id: Ia1cda5e117d7e569e8891bd7c3d03bd6148a0c27
Closes-Bug: #1374902
|
|\ \ |
|
| |/
| |
| |
| | |
Change-Id: Ib2e988b23dc5f264c07342343c848308de8eb2c9
|
|\ \ |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
When using serial console we are expecting using a plain websocket
so we should to return the 'ws://' scheme as the default one, also
it makes no sense to handle webserving content since currently
nothing has been implemented in favor of this case.
DocImpact: The 'base_url' option serial_console has been updated
Closes-Bug: #1373950
Change-Id: I0d0e4f7060febec5e0a357cd3e8c05486f2afaa5
|
|\ \ \ |
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The Ironic virt driver was previously returning from power_on,
power_off, and reboot calls immediately after making the request to
Ironic -- but before Ironic had necessarily acted upon it. This could
lead to Nova acting as though the state change was complete before it
actually was.
This patch introduces a new private method which polls Ironic until it
signals that the power state change is complete. With this, the
power_on, power_off, and reboot methods wait for the action to complete
before returning to ComputeManager.
Closes-bug: #1310135
Change-Id: I0bbdd8bc8807a0df43a420825f497842b387c4d5
|
|\ \ \ \ |
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
During the VM resizing, before VM arrive RESIZED state, driver
migrate_disk_and_power_off will initially rename orginal vm
'uuid' to be 'uuid-orig' and clone a new vm with 'uuid' name.
When deletion VM is triggered at this time window, it wouldn't
be able to delete the VM uuid-orig in VCenter and so cause VM leak.
As VM task state will be set to 'deleting' and it can not be used to
determine the resize migrating/migrated state, this fix will
attempt to delete orig VM within destroy phase.
Change-Id: I7598afbf0dc3c527471af34224003d28e64daaff
Closes-Bug: #1359138
|
|\ \ \ \ \ |
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
When spawning some instances, nova VMware driver could have a race condition
in VNC port allocation. This fix ensures that the lock is done on the
actual setting in the VM configuration spec.
Co-authored-by: Marcio Roberto Starke <marcio.starke@serpro.gov.br>
Change-Id: I70fab021bbf2df418df53e5f47e19cf16dbe45ac
Closes-bug: #1357372
|
|\ \ \ \ \ \ |
|
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
This commit modifies the reserve_block_device_name method to return the
bdm object, when the corresponding keyword argument is True. This
ensures the correct bdm is destroyed if the attach fails. Currently the
code assumes only one bdm per volume and so retrieving it can cause the
incorrect db entry to be returned.
Change-Id: I22a6db76d2044331d1a846eb4b6d7338c50270e2
Closes-Bug: #1349888
|
|\ \ \ \ \ \ \
| | |_|_|_|/ /
| |/| | | | /
| |_|_|_|_|/
|/| | | | | |
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
As part of the blueprint serial-ports when connecting to
the proxy with a plain websocket, make Nova able to retrieve
the token from the request path.
Closes-Bug: #1373950
Change-Id: I0b0b23c954dbb812acfa7616696445c0e167d2c8
|
|\ \ \ \ \ \
| |_|_|_|/ /
|/| | | | | |
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
While fixing issues in Cinder for bug 1367454 I discovered
that Nova doesn't have any good path test cases to make
sure that settings to osapi_compute_workers is being handled
as expected.
This commit adds those test cases.
Change-Id: I161adf9e93bd7ff1389fdbc8f712ae67ed98b613
|
|\ \ \ \ \ \ |
|
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
The detach_interface method is catching a libvirtError and logging an
error message but it doesn't include the contents of the root
libvirtError, so log that information as well for debugging.
Closes-Bug: #1374140
Change-Id: I042c2956c7393a2ee1ccd026421c8df537aa1a87
|
|\ \ \ \ \ \ \ |
|
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | | |
The current implementation of nova.tests.virt.test_virt_drivers
indicates NotImplementedError should be caught and logged to
dynamically build the hypervisor support matrix. This is not a
realistic / optimal approach for building the support matrix and
thus this patch updates the code comments accordingly for
clarity. This patch also modifies the logging statement outputted
in such scenarios to better clarify which test case method is
executing.
Partial-Bug: 1209421
Change-Id: I5d80cdc06207071548fdb5f6852ca96fa8d79776
|
|\ \ \ \ \ \ \ \ |
|
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | | |
This looks like a legacy. _db_content['files'] is always initialised,
and checking it for None distracts from what is actually being tested.
This was also the only use of NoFilesFound, so we remove it entirely.
Change-Id: Ided6c71b68a654476512016ddf886e895d75e571
|
|\ \ \ \ \ \ \ \ \ |
|
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
This commit removes the duplicate entry in the gitignore file for the
sample config file.
Closes-Bug: #1375432
Change-Id: Ied0fb5b38a616bbde871dfc690ccccfd46c94bde
|
|\ \ \ \ \ \ \ \ \ \
| |/ / / / / / / / /
|/| | | | | | | | | |
|
| | |_|_|_|_|_|_|/
| |/| | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | | |
If instance path is shared between compute nodes (e.g. over NFS) and image
backend uses files placed in the instance directory (e.g. Raw, Qcow2), live
migration should consider block storage of the instances shared.
Change-Id: I913c57152974562765502ab57be63ccb6dfe5208
Closes-Bug: #1351002
|
|\ \ \ \ \ \ \ \ \ |
|
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
The syntax suggested for adding hooks to nova via setup.py entrypoints is:
entry_points = {
'nova.hooks': [
'resize_hook': your_package.hooks.YourHookClass,
]
},
But this is incorrect. The proper syntax for an entry_points entry is:
entry_points = {
'nova.hooks': [
'resize_hook=your_package.hooks:YourHookClass',
]
},
Change-Id: Id7bc6891dc542d4317e82e15300c7293eb0d17a0
Closes-bug: #1374666
|