| Commit message (Collapse) | Author | Age | Files | Lines |
|\ |
|
| |
| |
| |
| |
| | |
Change-Id: Ic421350c4c5f8278ac020ee6a521026960e88634
Closes-Bug: #1319637
|
|\ \ |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Remove the 'fake' and 'ssh' drivers, since they should not show up in
production, and it is better to have production-oriented sane defaults.
Updates the option doc string with a lengthy description.
Updates the developer documentation with a link to the wiki page,
where a list of CI-tested drivers will be maintained.
Change-Id: I1151ace22752efdf6a8a43e279a57d728ea6cce2
|
|\ \ \ |
|
| |/ /
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
In order to avoid random iscsiadm errors this patch adds retry logic
to the iscsiadm commands. All iscsiadm commands will be attempted
5 times before failing.
Closes-Bug: #1327392
Change-Id: I971db541efbaaaf76bb6266570d7606d5b8285e7
|
|\ \ \ |
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
When there is no db engine configured for the migration tests,
the test_walk_versions test was passing, rather than skipping,
because it contained a for-loop and no check for the length of the
sequence being iterated over.
This patch checks the length of the configured engines and
properly skips the test when there are no engines.
Change-Id: I8a5c0a66f65a99c5169b251be02a8fbe6c727641
Closes-bug: #1327397
|
|\ \ \ \ |
|
| |/ / /
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Mock the python-seamicroclient library from within ironic/tests/drivers
so that test_seamicro.py can import ironic/drivers/modules/seamicro.py
and run the unit tests on it, even when the external library is not
present.
Change-Id: I4f2cd75b2c64a5a280a5d45d5cea058fd432815b
Closes-bug: #1312321
|
|\ \ \ \ |
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
The node parameter is redundant since the node object is an attribute
of the task object.
Partial-Bug: #1312632
Change-Id: Ic7bf898b7d881e36188511d75837e450f29f066c
|
|\ \ \ \ \
| |/ / / / |
|
| | |_|/
| |/| |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The multi-node locking is not used and there's no plan to use it.
Removing it allows us to clean up a lot of code.
The TaskManager tests are refactored to not use the DB.
Removed 'node' from get_node_mac_addresses() in driver/utils as it's
no longer needed.
Partial-Bug: 1312632
Change-Id: Ie3577ba150df6c84fc34a4b08775e5011f8925cb
|
|\ \ \ \ |
|
| |/ / /
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
We currently poll for success, but physical machines can take a while
to respond to IPMI etc. The current defaults reliably fail to complete
with 5 workers (which is the desired default). Until we have event
based logic from Ironic -> Nova we need to try harder, or be
unreliable. 60 retries of 2 seconds each means 2 minutes before a
truely hard fail (as opposed to queue backlog) is reported. Note that
deploys can starve deletes at the moment, so this fix is not enough to
stop spurious reports of problems.
Change-Id: I917a4430e5e77936deb82343ff057b9adc9822ec
Partial-Bug: #1319858
|
|\ \ \ \
| | |_|/
| |/| | |
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Attempting to delete an iSCSI target immediately after the initiator has
logged out of the session can sometimes cause iscsiadm's database update
to fail. This allows the delete command to be retried up to 5 times on
failure. We also accept exit code 21 (no records found) as success.
Closes-bug: #1324670
Change-Id: I064fcd01e3bd713b0ec19cc1fdb562ca7fd8b666
|
|\ \ \ \ |
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
After the server accepts a request (202) it now returns an URI in the
HTTP Location header so that clients will know how to track the status
of the request.
Closes-Bug: #1317044
Change-Id: I5286ed762426dea02a6513c0e31cbc9a8023f803
|
|\ \ \ \ \
| |_|_|/ /
|/| | | | |
|
| | |_|/
| |/| |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
This patch reworks the logic around make_partitions when rebuilding an
instance with preserve_ephemeral set to true. The reworked logic will
no longer wipe the current partitions when rebuilding an instance with
preserve_ephemeral set.
Partial-Bug: 1317647
Change-Id: I15fc0d6323698eaa4c30b8aaa7fb740f51c035bc
|
|\ \ \ \ |
|
| |/ / /
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
list_instance_uuid() was dependent on list_instances(). The latter
was fixed to return the correct results but the former was not,
causing nova-compute start up failures. This updates it accordingly
to return the expected output via a client API call.
Change-Id: I27651ca01dbdcc95ca729da7dbbeafdc616eec4e
Closes-bug: #1326988
|
|\ \ \ \
| | |/ /
| |/| | |
|
| |/ /
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
To provide information about the location of a newly created resource
(201) the Ironic API now returns the URI for the new object in the HTTP
Location header.
A new function build_url() was created in the link.py by factoring out
part of the the Link.make_link() method so that we can reuse that part
of the code to build a URL without having to create a new Link object.
Partial-Bug: #1317044
Change-Id: I88b033d446aaed93fc90ab6497578d5cd729dea8
|
|\ \ \ |
|
| |/ /
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The logging message of _build_pxe_config method in
ironic/drivers/modules/pxe.py is like below:
Building PXE config for deployment 4.
"4" is a ID of the node, but users don't know ID of the node
even if using CLI or API.
Users can know only uuid of the node.
So that, uuid of the node should be written in the logging message.
This patch makes identify node by uuid not by id.
Change-Id: Ida220a97d5fc8f6c5adc5b5acf239bbee14ff847
Closes-bug: #1327085
|
|\ \ \ |
|
| |/ /
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The module which contains the create_engine() method is the sqlalchemy,
not slqalchemy.api.
Change-Id: I0b9e2f5b25f0c8ea9f7e55d863fa170cc9c09564
Partial-Bug: #1327225
|
|\ \ \ |
|
| |/ /
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This patch adds a check in the PXE driver validate() method to check if
the requested image exist in Glance and if it contains the 'kernel_id'
and 'ramdisk_id' properties set which are needed for the PXE boot.
Closes-Bug: #1324840
Change-Id: I42bf7fcb27286afb3db0fda2a3a09ea49c77496b
|
|\ \ \
| |_|/
|/| | |
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The _unprovision function used a class member to track the number
of calls to ironic to check the provision state. When more than one
delete is happening concurrently, the deletes will both add to the
same counter. This commit fixes this issue to use a local counter.
Partial-Bug: #1326364
Change-Id: I2552332445fe0e614970ec28eddf7c1ae94e34be
|
|\ \ |
|
| |/
| |
| |
| |
| |
| |
| |
| |
| | |
If deploy or tear_down raise an exception, an error is set on the Node
and visible in the API, but nothing is logged. This change adds
logging in the ConductorManager of exceptions raised from either
method.
Change-Id: I17a9328770b6fc9fce167a189d9692b5c4c2687f
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
A good idea because it makes the code easier to read to get rid of
the `with` using mock.patch and mock.patch.object in favor decorators,
where possible in the tests.
This patch affects Nova driver tests.
Change-Id: Ifd2e4d36c3cbcf82cb9b83c375ba0a3e98019adf
|
|/
|
|
|
|
|
|
| |
To use these filters out of tree right now, you have to specify them in
nova's available_scheduler filters by full path. But in order for the
modules to be found and loaded, you must have an __init__.py.
Change-Id: I4816d9dbe0e93056792fad48e6228ffdd531e346
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Previously the list_instances() method in the Nova Ironic driver was
returning a list of instances UUIDs instead of a list of instances names
causing various checks throughout nova.compute.manager to always return
False for Ironic.
Closes-Bug: #1309719
Change-Id: I0ed29911f180f99a7e2b1c00ddd6c43064f04f7f
|
|\ \
| |/ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
As the ClientWrapper class had no unittests some were added to test
the specific part of the code for this change. The FakeClient() and
associated classes from test_driver.py were moved to a common file to
make it easier to be imported and used by the new unittests.
Partial-bug: #1316549
Related-bug: #1309719
Change-Id: I11be25a6ce981e63027073d12fbf2922896f4693
|
|\ \ |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This patch fixes the doc about methods handling power state
to remove the parameter "node".
* validate
* get_power_state
* set_power_state
* reboot
Change-Id: I5c0425e1502f304b62566aa1e63823214888ef66
Partial-Bug: #1312632
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
A number of changes with keystoneclient breaks our tests. First, a
kwarg changed for cache.set(). It should be 'time' now, not 'timeout'.
Also: keystone middleware now looks for an 'expires' in the token.
Sample conf needed regenerated, also, due to changes.
Change-Id: I33c4f1bae5a3c89b47fd21f8db1b4a4d536ed369
Closes-bug: 1324655
|
|\ \ \ |
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The node parameter is redundant since the node object is an attribute
of the task object.
Partial-Bug: #1312632
Change-Id: I4baa8526768a250e0cf047f56322039157772376
|
|/ / /
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This patch sets the correct target versions for the RPC methods in
the RPC API. This is a first step towards being able to enable rolling
upgrades for Ironic.
Partial-Bug: #1317300
Change-Id: I67ce731731c38e48b4f3180a423cbc18a3728903
|
|\ \ \ |
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Currently, the NodeLocked exception's HTTP status code indicates a
server-side failure, resulting in WSME logging it as such, and this
is leading to random failures in our CI tests. While NodeLocked is
a valid issue which should (currently) be exposed to the client,
it is not a server-side error and does not warrant a 5xx status code.
This patch changes the HTTP status code for this exception to
409 CONFLICT, causing WSME to log it as a warning. This is not ideal,
but it resolves the immediate issue (random CI test failures).
A longer term solution is being drafted at:
https://review.openstack.org/94923
Change-Id: Ied9dcf9bbcb505177adc57d7d8fe06e4096ff488
Closes-Bug: #1321494
|