The `overcloud node unprovision` command will delete
ports by default. See: tripleoclient/v2/overcloud_node.py#L422
Let's do the same by default when using `overcloud delete` with
the `--baremtal-deployment` option.
Change-Id: I881fa2cbb6612e10dbe85a800f38c6922fc0a92b
When we check for bridge mapping. If the user
has provided the Name of the node, rather than
the UUID. Then we will fail to lookup the agent.
This change ensures that we lookup the agent using
the nodes UUID rather than the name, regardless of
what the user has input in the provide command.
Change-Id: Iceea8fbb3053aa0af7ba46bec43dcedb3e8801b8
... because the current master should be compatible with the Zed
release of the other projects. This effectively removes testing on
Python 3.6 and only Python 3.8 and 3.9 will be tested.
Python 3.6 and 3.7 are removed form classifiers because these are no
longer supported.
Related-Bug: #1974244
Change-Id: Ibafdb84e5573d74c02874dc0531a6645195a8b5f
The output cli switch was being ignored, this change will ensure that if
a user provides an output path the export function will operatate as
expected.
Change-Id: If53fa783c95d9d1031d04763f85278bea7e44b30
Signed-off-by: Kevin Carter <kecarter@redhat.com>
ssh_servers takes a list of strings. This change
ensures that each item in the list is a string
instead of type Host being returned from
InventoryManager.
Resolves: rhbz#2085492
Change-Id: Ic3dc6304287361e83e91ad05aa480ff874011f1b
This patch fixes an issue on process daemons that happened because
of the related function is called with a wrong name.
Change-Id: I4ee52fe91317d1a22bbd8c6f77e507dfae71c46b
The `check_deploy_backups` makes calls to `stat` which can potentially
lead to I/O error in testing scenarios when the relevant file is
removed after path retrieval but before `stat` is called.
This can lead to tox test failure, in both CI and local environments.
Mocking the utils.check_deploy_backups in tests where it is indirectly
called should alleviate the problem.
Long term solution would be mocking all built in functions performing I/O calls by default.
Closes-Bug: #1969425
Signed-off-by: Jiri Podivin <jpodivin@redhat.com>
Change-Id: I96d4bfce84ffe36d476e3383ee264cea6fd93c24
This removes tripleo-build-containers-jobs as this only contains
centos-7 jobs [1] and they are being removed with [2].
[1] 15761b77d9/zuul.d/build-containers.yaml (L119-L127)
[2] I70d2e5dbe995bc8dfe06817d6460f3f2f3732e7e
Change-Id: Idcab9a34ae7b5f20702d9a616173dd25d3894751
The validator run subcommand which was previously without
any unit test coverage now has it's basic operation mode tested.
Assertions are placed on calls and return values of the parser.
The existing test structure was improved with enabling of autospec
on mocked API.
Closes-Bug: #1970387
Signed-off-by: Jiri Podivin <jpodivin@redhat.com>
Change-Id: I008ac81cd803adbd08a9509902e46967f8b7d2e4
This service is now deprecated, and everything is managed in
tripleo-ansible/tripleo_firewall automatically.
We keep the _generate_masquerade_networks() method, since it adds some
intelligence and prevent the need for the user to actually set it -
therefore, we keep the automation based on the
undercloud.conf/<subnet>/masquerade=true
Change-Id: I09744be1f4766a2a8241bfa79095fa97ea82c602
Add new parameters --ceph-client-username, --ceph-client-key,
--skip-cephx-keys, and --ansible-extra-vars to 'openstack
overcloud ceph deploy' so that the user may create pools
and cephx keys for OpenStack to access them on the deployed
Ceph cluster. This is optional as 'openstack overcloud deploy'
will create the pools/cephx keys later, however the same
Ansible can now be called to do it earlier.
The --ansible-extra-vars option can override any variable
in tripleo-ansible when 'openstack overcloud ceph deploy'
is called. If the tripleo_cephadm_pools variable is set,
then cephx keys will be generated and exported to the
--output deployed_ceph.yaml. The user may also pass their
own cephx key and client username.
Depends-On: Ib87a8a5a550676b32302c0cc9c447f8b9fe3a149
Change-Id: Ic37d7123c30b559b1ff41e1d93ce7c7673b85a92
We changed the unprovision confirmation file structure to
include pre-provisioned nodes[1]. This fixes node delete
for that.
[1] Id377b2c7ed973a7e2365b33dba2c1d59a59e26b9
Change-Id: Id5b5e257679ce7fca26e1ebf272988eaab2d4d53
We should update the deployment status during config
generation using ephemeral heat stack.
As we cleanup the heat/stack at the end, there is no way
to know if the last config generation has failed (no stack).
Also, earlier deployment_status used to check for stack status.
Change-Id: I8b358444b1ce2b31c40a0994ba955c7baab9e6ac
Remove support for using an undercloud installed Heat, as it no longer
needs to be supported.
Change-Id: I39150aa6771a32318ad2cbcd50b6dddcd97b4663
Signed-off-by: James Slagle <jslagle@redhat.com>
Ephemeral Heat interaction was spread across several areas of the
overcloud deploy, including the entire config-download execution. To
reduce the amount of time that ephemeral Heat needs
to stay running, consolidate all the Heat interaction to a single
try/finally block. The overcloud export is moved earlier in the
execution since it no longer requires Heat API access. This also makes
the shutdown of ephemeral Heat and database backup in just a single
location.
Change-Id: Ia424b2eb251b2d31a97256b5e8882c70058d6a0b
Signed-off-by: James Slagle <jslagle@redhat.com>
Update the overcloud export command to work without a dependency on Heat
API. The export data is updated to create the export data based solely
on saved stack output data and the config-download generated
overcloud.json so that Heat is not needed to generate the export at all.
Change-Id: I0bd4b4a0e33184e0f4d731899138bf4063f3cdac
Signed-off-by: James Slagle <jslagle@redhat.com>
This command was still attempting to query Heat for the stack, which
would not work with ephemeral Heat.
Depends-On: I28c21f923baf75fbc3be501552c10a5bb5384a09
Change-Id: I176bc05bf588f12c543273fa1d5e357ea819fcd3
Signed-off-by: James Slagle <jslagle@redhat.com>