The 'openstackclient.compute.v2.server' module is getting rather large.
The server migration commands don't need to be in there. Move them.
Change-Id: I8b2600cfd9f8e37d3093c52c7222d85e84e7fc89
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
The nova API doesn't allow you to retrieve migration records by UUID,
only ID. This is confusing. Work around it by listing records and
filtering this list.
Change-Id: I932c9c70420e85056509513e005bb78168e70611
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
Why limit a user to preset ssh arguments? Capture them all and send
them along to ssh to deal with. This allows users to use the full range of
ssh arguments, including specifying a command to run on the
instance. For example:
openstack server ssh -4 upg -- -l cirros -i ~/id_rsa_upg "date; uptime"
SSH arguments that openstackclient currently mirrors are deprecated
except for -4 and -6, as they are useful for retrieving the correct
instance IP.
Change-Id: Ia50786d5eee52688e180550fe16aeb8af610154b
Co-authored-by: Stephen Finucane <stephen@that.guru>
argparse allows you to specify multiple options for a given argument
when declaring the argument. For some reason, we weren't doing this for
the 'server ssh' command. There's no apparent reason for doing things
this way and it's been that way since the beginning (2013) so let's not
do that.
We also add unit tests since they were missing and should exist.
Change-Id: I67a9e6516d7057266210cd4083e9ddeb1cfaa5de
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
The compute API provides this information to us. We might as well use
it.
Change-Id: I5608fa80745975ce49712718452cfe296c0f64d2
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
The nova API we're using to delete a server volume attachment needs to
be handed a volume, not a volume attachment.
Also make sure that we create an error if the volume isn't actually
attached to the server.
Signed-off-by: Dr. Jens Harbott <harbott@osism.tech>
Co-authored-by: Stephen Finucane <sfinucan@redhat.com>
Change-Id: I12abd3787ea47acb4da282d00fdc1989405a0564
Fix the issue where the flavor name was empty in server list output.
This requires somewhat invasive unit test changes to reflect the changed
API response from the server, but this has the upside of meaning we
don't need new tests since what we have validates things.
Also drop the flavor ID column as it is removed from the compute API.
Change-Id: Ica3320242a38901c1180b2b29109c9474366fde0
Signed-off-by: Khomesh Thakre <khomeshthakre24@gmail.com>
Story: 2008257
Task: 41113
This has no impact on the end result, but it should make fixing issues
introduced by API microversion 2.69 a little easier.
Change-Id: I7d70eac8aa1a6197ed05a49f071e6899ec219c03
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
Switch the server suspend and server resume commands from novaclient to
SDK. Use the SDK versions of test fakes to support fake Server
resources.
Change-Id: Idd0b4f13fab0f238e42844a7d759538bbda24f68
Switch the server pause and server unpause commands from novaclient to
SDK. Use the SDK versions of test fakes to support fake Server
resources.
Change-Id: Id626f06f3d7edd44b306b7fc7b9b00d04af09621
File tests.unit.volume.v2.fakes is modified to provide sdk volume fakes.
File tests.unit.compute.v2.fakes is modified to provide sdk volume
attachment fakes. For test, setup_sdk_volumes_mock() method is created
so that volumes are created in similar way as servers are created.
Change-Id: I290ba83b6ba27a1377ab73fd0ae06ecced25efd1
The old novaclient.v2.server.Server.interface_attach() method is
replaced with proxy.create_server_interface().
In swargs, 'net_id' and 'port_id' are mutual-exclusive, if one of
them is given with value, the other one cannot be None, as the
API would responde with 400 (None is not string).
In unit test, temporary method 'setup_sdk_servers_mock' is added,
because other tests are still using the old 'setup_servers_mock'.
Functional tests are added. Releasenote is generated.
Change-Id: I9899f0509febc5143560a1859ae6344d0a6d1427
The API behind the 'server add volume' command returns details of the
created volume attachment, however, we were dropping these results
rather than displaying them to the user. Correct this.
Change-Id: I3f7e121220d29422ccf4e6940de2f28bb8496c83
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
We should be inheriting from 'ShowOne'. Failure to do so results in a
tuple being dumped to the screen. Not what we intended.
While we're here, we update the docstring of this command to clarify the
command's intent. Nova does not provide an API to retrieve an individual
migration record for a cold migration or completed live migration. As
such, the 'server migration show' command only works for in-progress
live-migrations.
Change-Id: I2e2fe3da7d642b9e8e3d930603dcde178cd68cde
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
Story: 2009658
Task: 43837
Due to a small logic error, we were emitting a warning about a
deprecated option when the user tried to live migrate an instance using
microversion 2.25 even though the user hadn't actually set that option.
Correct this.
Change-Id: Ib61e817bd4ced9b5533e7c7f9d8f0b45fe81c211
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
Story: 2009657
Task: 43836
The nova API expects the 'tags' and 'not-tags' filters of the 'GET
/servers' (list servers) API to be a CSV string [1]:
tags (Optional)
A list of tags to filter the server list by. Servers that match all
tags in this list will be returned. Boolean expression in this case
is 't1 AND t2'. Tags in query must be separated by comma.
New in version 2.26
not-tags (Optional)
A list of tags to filter the server list by. Servers that don’t
match all tags in this list will be returned. Boolean expression in
this case is 'NOT (t1 AND t2)'. Tags in query must be separated by
comma.
New in version 2.26
We were instead providing a Python list, which was simply being URL
encoded. Correct this.
[1] https://docs.openstack.org/api-ref/compute/?expanded=list-servers-detail#list-servers
Change-Id: Ie0251a0dccdf3385089e5bbaedf646a5e928cc48
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
Closes-Bug: #1946816
this already exists for server rebuild, but was missing for server
create.
This option is supported from Compute API version >= 2.63, and is only
available for servers booted directly from images (not from volumes,
not from snapshots, and not from images first converted to volumes).
Additionally, this patch removes mentions of
OS_TRUSTED_IMAGE_CERTIFICATE_IDS env var from similar option help string
in server rebuild command as it is not actually implemented yet.
Change-Id: I4e9faea05c499bd91034d1d284c44fdcc8e18db5
Allow configuring hostname when creating a new server or updating or
rebuilding an existing server.
Change-Id: Ibe603eab78bbbec43605f56de62a20493b6aa93d
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
Depends-On: https://review.opendev.org/c/openstack/python-novaclient/+/806917
openstack server list -c "Created At" command doesn't work
because the wrong variable was used here.
When we receive resp data, Created At data is saved with
the name "created". But in "server.py", we append columns
as created_at. So it seems to print an empty table.
Story: 2009149
Task: 43112
Change-Id: I06de6903d5cc427a8b0fdcd168fec47192f4365b
The compute APIs are unfortunately inconsistent with regard to the
response parameters for migrations.
* GET /servers/{server_id}/migrations returns server_uuid
* GET /os-migrations returns instance_uuid
Because the 'Server UUID' column is being specified for parsing the
response from GET /os-migrations, it is always showing as an empty
string to users.
There are a few other mismatches between the column names and the REST
API response fields [1]:
* 'Old Flavor' vs 'old_instance_type_id'
* 'New Flavor' vs 'new_instance_type_id'
* 'Type' vs 'migration_type'
This adds a new list containing the REST API response field names to
pass to utils.get_item_properties so that the responses are correctly
parsed and the client output contains the response data instead of
empty strings.
Story: 2009078
Task: 42890
[1] https://docs.openstack.org/api-ref/compute/?expanded=list-migrations-detail#list-migrations
Change-Id: I8aab60619e0225047f6a1c31e44917ca8fcc799e
Manually changing the server state is a potentially dangerous operation
that should only be done under limited circumstances. It's also an
admin-only operation by default. Highlight both points.
Change-Id: Ifd8aec94937764202131ba8caf6b507caa76d7e9
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
Story: 2008549
Task: 41672
Password injection requires either hypervisor-support or an agent
running in the guest that will talk to the metadata service. It can be
disabled for a deployment using the '[api] enable_instance_password'
nova config option. Indicate this, albeit briefly.
Change-Id: Ief94ea07fc7ab6a487af972e8759ca6704d8f085
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
Update the help strings for these two arguments to indicate their
deprecated nature. This was previously flagged via a deprecation warning
but users would only see that if they were to run the command.
Change-Id: I31a5e27ac8bd2625a6073b54a51bf3e8d6126c8c
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
There are a couple of other (networking-related) options which accept
paths, none of which insist on a URI-style path. Let's just drop this
bit of complexity before we release the feature.
Change-Id: Ia7f781d82f3f4695b49b55a39abbb6e582cd879c
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
The syntax of the '--block-device' parameter is complex and easily
screwed up. Allow users to load a block device config from a file. For
example:
$ openstack server create ... --block-device file:///tmp/bdm.json ...
This should alleviate the pain that is BDMv2 somewhat.
No functional tests are provided since we already have tests for the CSV
style of passing parameters and the unit tests show that the net result
is the same.
Change-Id: I3e3299bbdbbb343863b4c14fb4d9196ff3e1698d
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
This mostly reuses the existing tests for '--block-device-mapping',
which can hopefully be removed at some point in the future.
This highlights two issues with the implementation of this option.
Firstly, the 'boot_index' parameter is not required so don't mandate it.
Secondly, and more significantly, we were defaulting the destination
type for the 'image' source type to 'local'. Nova only allows you to
attach a single image to local mapping [1], which means this default
would only make sense if you were expecting users to use the
'--block-device' option exclusively and omit the '--image' option. This
is the *less common* case so this is a bad default. Default instead to a
destination type of 'volume' like everything else, and require users
specifying '--block-device' alone to pass 'destination_type=local'
explicitly.
[1] https://github.com/openstack/nova/blob/c8a6f8d2e/nova/block_device.py#L193-L206
Change-Id: I1718be965f57c3bbdb8a14f3cfac967dd4c55b4d
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
The parameter isn't actually deprecated, since we need to support older
API microversion, however, we now emit an error if someone attempts to
boot a server with the wrong microversion. This would happen server-side
anyway since this parameter was removed entirely in API microversion
2.57.
Change-Id: I73864ccbf5bf181fecf505ca168c1a35a8b0af3a
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
These options are not supported from Nova API microversion 2.25 and
above. This can be a source of confusion. Start warning, with an eye on
erroring out in the future.
Change-Id: I53f27eb3e3c1a84d0d77a1672c008d0e8bb8536f
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
API microversion 2.25 introduced the 'block_migration=auto' value for
the os-migrateLive server action. This is a sensible default that we
should use, allowing users to avoid stating one of the
'--block-migration' or '--shared-migration' parameters explicitly.
While we're here, we take the opportunity to fix up some formatting in
the function, which is really rather messy.
Change-Id: Ieedc77d6dc3d4a3cd93b29672faa97dd4e8c1185
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
One of the last big gaps with novaclient. As noted in the release note,
the current '--block-device-mapping' format is based on the old BDM v1
format, even though it actually results in BDM v2-style requests to the
server. It's time to replace that.
Change-Id: If4eba38ccfb208ee186b90a0eec95e5fe6cf8415
Signed-off-by: Stephen Finucane <sfinucan@redhat.com>
