This no longer makes any sense since all network calls uses the SDK
proxy API now.
Change-Id: I0d57d7561b697270b24acd7b8ff62db73a83f49a
Signed-off-by: Stephen Finucane <stephenfin@redhat.com>
This was still using glanceclient-style calls despite us having switched
all image command to SDK way back in 5.2.0 (Ussuri). It therefore hasn't
worked since then, so we remove without deprecation.
Change-Id: Ibea79ea10bb272cba194debfe4bcbaeebf3875b7
Signed-off-by: Stephen Finucane <stephenfin@redhat.com>
Migrate the block storage aspects of this command to SDK. This means the
'extension list' command, like the 'availability zone list' command
previously, is now using SDK entirely.
While we're here, we also make some fixes to the unit tests for the
network and compute aspects of the command. While we migrated the
network and compute extension commands from neutronclient and novaclient
respectively some time back, we never fully updated the tests. Do this
now.
Change-Id: I631a6a09dfd9d614b1dd7b322dcee8490a52cc43
Signed-off-by: Stephen Finucane <stephenfin@redhat.com>
Depends-On: https://review.opendev.org/c/openstack/openstacksdk/+/885132
This is a bit of an odd one. Because this is a "common" command, there
are three clients in play here: a compute client, a block storage
client, and a networking client. The networking aspects of this command
are already using SDK though the tests were mistakenly using a fake
version of the legacy neutron client (albeit one with the correct SDK
properties). We were still using legacy clients for the other two.
Correct the networking tests and migrate the compute and block storage
aspects of the command.
Change-Id: I292e1a712bca97e1270e8a97606e02a8a44c2954
Signed-off-by: Stephen Finucane <stephenfin@redhat.com>
The --flavor-id option is added to the router create command. The
flavor_id attribute has been supported by the Neutron API for router
POST operations since a long time ago [0].
Partial-Bug: #2020823
[0] https://review.opendev.org/c/openstack/neutron-lib/+/494289
Change-Id: If88499533a92c09e67b0827d1d64156682a3bb8e
The existing 3.9 is based on debian buster and isn't getting security
updates ... or any updates really. This change updates to a base
container that will get updates for a while.
I picked 3.10 rather than 3.11 as 3.10 is part of the bobcat[1]
[1] https://governance.openstack.org/tc/reference/runtimes/2023.2.html
Change-Id: I606389a81ab189b1abf2c700acf5c5543a09165a
Took me a bit to figure out how to do token auth today. Document it.
Change-Id: I3edce59efd5ca5a6e31eef8c7cf3cf255234c82e
Signed-off-by: Stephen Finucane <stephenfin@redhat.com>
This document is growing a bit too big now.
Change-Id: Iadf7f28d2307d1598e378bd49a09e75e1235d779
Signed-off-by: Stephen Finucane <stephenfin@redhat.com>
This patch will add ``image metadef resource type list`` using
SDK, output matches glance client output.
Change-Id: I29850abdbb06d5a4eaf8a9a60f565b46f4009f85
These mess with quotas on the main project, which can have knock-on
effects for other tests. Use a new project.
Change-Id: I687ec3a0985eab4d01c918dae73d2fbdc7ac25fa
Signed-off-by: Stephen Finucane <stephenfin@redhat.com>
The help text for this arg says 'repeat option to delete multiple tags'.
Fix to reflect what docs says.
Also add missing test for this argument.
Change-Id: Ib1069ce7a441c1ff10d2dca05095eb6bf53e7fb6
This patch modifies the access rule commands to use only the resource
ID. The previous logic incorrectly assumed that access rules have a
"name" property, which resulted in unexpected behaviors.
For example, "access rule delete {non-existent-id}" now results in a
"not found" error instead of sometimes deleting an unrelated rule.
Story: 2010775
Task: 48163
Change-Id: Ib5c3b7f86acf1dfe7cc76dfa99fa4c118388bd71
This is more trouble than it's worth now. Black takes care of this for
us. I wish I'd decided this before the blackify series :(
Change-Id: Ifee786c96b6f0bcd96008fed2b3ed3257d02c07c
Signed-off-by: Stephen Finucane <stephenfin@redhat.com>
Introduce the hints port attribute that allows passing in backend
specific hints mainly to allow backend specific performance tuning.
Enable:
openstack port create --hint ALIAS=VALUE
openstack port set --hint ALIAS=VALUE
openstack port unset --hints
Required neutron extension:
port-hints
port-hint-ovs-tx-steering
Valid hint aliases and values:
ovs-tx-steering=hash
ovs-tx-steering=thread
The same hints in JSON format, as expected by the Neutron API:
{"openvswitch": {"other_config": {"tx-steering": "hash"}}}
{"openvswitch": {"other_config": {"tx-steering": "thread"}}}
Change-Id: I4c7142909b1e4fb26fc77ad9ba08ec994cc450b2
Depends-On: https://review.opendev.org/c/openstack/neutron/+/873113
Partial-Bug: #1990842
Related-Change (server side): https://review.opendev.org/c/openstack/neutron/+/873113
Related-Change (spec): https://review.opendev.org/c/openstack/neutron-specs/+/862133
These are already set in the 'TestServer' super class.
Change-Id: I90cfd9ff004c59cadc3e760eb120a460687701fe
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
Dropping the use of 'AddFixedIp' server action entirely in favour of
creating an interface with a fixed IP using the 'os-interface' API. The
reason for doing this is because this mechanism allows us to request an
explicit fixed IP.
Adds back support for using the --fixed-ip-address argument for users
who use older nova-api versions.
Change-Id: Ieac4b0fbcb38fe7207eaaa6a79e0c6770ee11d80
There are a few outstanding commands. Migrate them in one fell swoop. We
do lose a small bit of information about commands that are only
supported in block storage API version 2.0 or later, but who cares about
block storage API version 1.0 these days?
On a related note, we also use the 'openstack.volume.v3' entrypoint
rather than the v2 one since that's more relevant nowadays and the code
is pretty much identical.
Change-Id: Ide9d65596847d3f855981ff4ec57eeaeea6667a7
Signed-off-by: Stephen Finucane <stephenfin@redhat.com>
Supporting "--no-property" option will apply user a convenient way to
clean all properties of volume qos in a short command. The patch adds
"--no-property" option in "volume qos set" command and update related
test cases and docs.
Change-Id: I1fb5b4f0a923bbf557a3af3f63809bde9e84ffd4
We also rename the test file and fixup some of the tests.
Change-Id: I3731255648dc91c023a1390c3b37e68b6608f850
Co-authored-by: Stephen Finucane <stephenfin@redhat.com>
Story: 2008478
Task: 41518
A mistake was introduced during the conversion from novaclient to SDK in
change I5ebfa6b2468d5f20b99ea0eab1aea9377be09b8c. Fix the issue and add
functional tests to prevent it being reintroduced.
Change-Id: I6b314eab31bcf452e88b8b6a239ac2e296497cb9
Signed-off-by: Stephen Finucane <stephenfin@redhat.com>
Story: 2010750
Task: 48004
currently the command is waiting only for ACTIVE server status,
but if the server was SHUTOFF before, it will be SHUTOFF after
rebuild as well, so the command is stuck in waiting forever.
Additionally, we now also pre-validate the server status on client side,
and raise an error if the server to be rebuilt is not in ACTIVE, ERROR
or SHUTOFF state.
Change-Id: If90a4bbba9a7ecd972f8b594c52fee4f75a0ae5e
Co-Authored-By: Oleksiy Molchanov <omolchanov@mirantis.com>
Story: 2010751
Task: 48005
We had not migrated a number of tests to use SDK objects instead of fake
novaclient-like objects when migrating the commands themselves. Address
this now.
Change-Id: Ib0da07fd9d793968b111986bd36a6d4311469d4e
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
This will make future changes slightly less complex.
Change-Id: Ia69fd9d05796cce5ed48a8d12902cb54caec9779
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
Since our pep8 tox env now runs pre-commit, we can ensure that black
styling will be maintained.
Change-Id: Ifad1d67a83c4979b4f4de71de0124613d1133560
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
