Age | Commit message (Collapse) | Author | Files | Lines |
|
1. Add the type: NodePort in dashboard service. the default is ClustIP,
which cannot access from outside.
2. Print the url ,user, password for user to access dashboard.
3. configure the kubectl CLI in opnfv host.
Change-Id: I6cb6e6f7547412139ece0c40a85de67a9edce0ef
Signed-off-by: wutianwei <wutianwei1@huawei.com>
|
|
OPNFV scenarios are forced to support at least two versions of OSA.
If there are API changes between two supported versions of OpenStack,
the testing code must have two versions. However, without knowing what
OpenStack version is deployed, it is impossible to automate the
decision of which testing code to use
Change-Id: I3ca8c6f33c05826a5d69417da774b7218bbd59e2
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
|
|
|
|
|
|
|
|
The configure-targethost playbook is configuring all the OSA hosts so
we can simplify all that by introducing a new 'openstack' host group
with the controller and the compute nodes and configure all of them
in parallel.
Change-Id: I21f8709e7649b041575dab45fc4e070aa01e0a6f
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
There is no reference on the actual upstream problem so lets assume
this is fixed and drop the workaround we had locally.
Change-Id: I3f71ea2e6fdc67e342f014e572f74760a78d3816
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The DEFAULT_XCI_TEST value wasn't checked and it could actually contain
anything leading to all sort of failures so we need to check it against
well known values.
Change-Id: Ib12c944adebfbcb3097b42af98d2bacefbd8374c
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
All clean VM operations should be performed within the root directory of
the releng-xci repo so make sure we are there.
Change-Id: I96f9d91439210faa1ac7f0e43db3992776293e9e
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The XCI hosts may need some of the XCI env variables to perform
certain tasks so dump the XCI environment to /root/xci.env file and
make that available to every host on the deployment.
Change-Id: I286a01ca7da8ff206438c261798167f5e4daf7f2
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
There is probably no good reason anymore to be so strict with file
inclusions in the roles directory so drop the .gitignore file.
Change-Id: I84569c8c82c4aeb5da6c6e39e02ff514566e5213
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The qemu-kvm package was duplicated for opensuse distro in the
start-new-vm.sh script. It doesn't affect its functionality but
reduce the list of additional packages to install.
Change-Id: Id5f24793295d7ccd9b19b64a07b24a1150a780aa
Signed-off-by: Victor Morales <victor.morales@intel.com>
|
|
Change-Id: If53b51fc26d642afa39c5e618afcd23fab395acc
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
|
|
|
|
|
|
|
|
Pipefail should be sufficient to determine the failure so this
change gets rid of grepping the logs for failure and unreachable
to leave the evaluation to bash itself.
Change-Id: I0fcd6016ddc19eb11b3eb2402347032a11c5a32a
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
|
|
|
|
Those roles were too small and only used as part of the host
bootstrapping process. As such, we merge them into a common
'bootstrap-host' role which can be used to prepare the hosts
after the initial deployment
Change-Id: Ifc84cf40b98ced91b31aae699dc28e9642380550
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Ansible gets upset if we use the same name for a host and a group so
add the OPNFV host to the deployment group. This fixes the following
warning:
[WARNING]: Found both group and host with same name: opnfv
Change-Id: Idbaff712d40e8234721332aee6d10d5ba44c62ce
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
'allow-pass-ansible-arguments-force-verify'
* changes:
xci: OSA: Fix warning about missing inventory file
xci: Pass the XCI_PATH variable to all Ansible calls
xci: Rename XCI_ANSIBLE_VERBOSITY to XCI_ANSIBLE_PARAMS
|
|
CI sometimes fails with the following error
"msg": "fatal: unable to access 'https://git.openstack.org/openstack/dragonflow/':
Failed to connect to git.openstack.org port 443: Connection timed out"
We assume that the upstream OpenStack mirrors may get very busy at times
or the connectivity between the CI and the OpenStack repos is not very
stable. As such, lets switch to github.com for the majority of OSA
repositories such as the roles and the services which may improve the
situation.
Change-Id: Ia4668d692077a785c76adeda128eb4cf9f4516a7
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Pipefail should be sufficient to determine the failure so this
change gets rid of grepping the logs for failure and unreachable
to leave the evaluation to bash itself.
Change-Id: Ie928438c1503b086159276af4308f5549b58cb71
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
The VM nodes are named using OpenStack terminology but since we
now have Kubernetes scenario, it is appropriate to name the nodes
accordingly.
Apart from that, we normally log the version of OSA which is not
correct for kubernetes deployments. Logging Kubespray version would
be correct for this scenario.
Change-Id: I3964574c8592450f9fa3126f19e27bb78fd9d174
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
|
|
|
|
There is no good reason why we should pin the Ansible version without
the ability to override it. OpenStack Ansible is doing a great work
testing Ansible releases as early as possible so we could benefit from
that work and use whatever Ansible our pinned OSA uses.
Change-Id: I4677dbefc68f7140417ece34d013ca9674baab43
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Change-Id: Ie196d1df537d09f0f91e43ab5e0305a45d543815
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: I330bc036f901d4ba61bc94ee6e085cadf54b4d8b
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
This is similar to 346079ea1b8dbda0c5e282c18f30cbac7e907d1e
("xci: Fix warning about missing inventory file")
This playbook only affects localhost so make it explicit. Fixes:
[WARNING]: Host file not found: inventory
[WARNING]: provided hosts list is empty, only localhost is available
Change-Id: If8548a36498ea3191de33cff94acbaee6ed04ce3
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The XCI_PATH variable is used by various playbooks to find the root
directory of the XCI repository so it's much cleaner to pass it on every
Ansible call.
Change-Id: Ifc0b77fa767e48bd1dc6daa44c6251c02983fd4d
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
We don't quite need a specific env variable just for Ansible
verbosity so we can rename this variable to make it clear that
it can be used to pass any Ansible option to XCI.
Change-Id: Ie20517d4b563bfc6daeb27848168d36da7014cee
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The upstream pw-token-gen tool doesn't need python-crypto anymore since
e9f957861b4160640f6debb2b939084ec43b43b2 ("Make pw-token-gen.py more
random") so we no longer need to install that package.
Change-Id: Ib53f246db999ff8ecfed2e3f62143c780c483fbd
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
|
|
|
|
* changes:
xci: OSA: Ensure proper constrains are applied to pip installations
xci: Move functest required packages to the functest role
|
|
|
|
We need to make sure that the local package cache is updated before we
try to install packages. Moreover, from time to time, something in the
repos may be broken so we don't want to break the CI job. As such, make
that step non-fatal as simply send the output to stdout so we can see
what's wrong.
Change-Id: I148e3b5435c604bb0d54339935255740c8ce07c0
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The package module gets upset if we pass it an empty string as a package
name
Mar 09 15:15:33 TASK [configure-network : Ensure networking packages are present] **************
Mar 09 15:15:49 fatal: [opnfv]: FAILED! => {"changed": false, "failed":
true, "msg": "No package matching '' found available, installed or
updated", "rc": 126, "results": ["iproute-3.10.0-87.el7.x86_64 providing
iproute is already installed", "No package matching '' found available,
installed or updated"]}
As such, lets pass an existing package for non-Debian distros to make
the module happy.
Change-Id: Ib1128057097193278cfe50f79daf35332b54e781
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
We need to make sure that the pip packages that we install are
compatible with the OSA components.
Change-Id: I87e80dc9b8fb862d9f9039d835b0908f752de3ca
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The docker packages that we install in the OPNFV VM are needed by
functest so add them to the related role.
Change-Id: I6ebe76fd030859f757d41ecf20c30ab76888ee9c
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
|
|
We no longer have a global inventory file and we don't need one right
now since the bootstrapping opearations happen on the localhost anyway
so we could be explicit and get rid of the following warning:
[WARNING]: Host file not found: inventory
[WARNING]: provided hosts list is empty, only localhost is available
Change-Id: I5cfe7b0971397665cf9eae9c54985c44c4506449
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
the network"
|
|
The DIB variables only make sense on bifrost so they should live in the
bifrost provision script.
Change-Id: I05baca7b9c6889a0d401a9288998991324d270af
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Ansible some times loses connection with the host after we configure its
network interface. This shouldn't happen since the host is actually up
and running. As a result of which, we need to fire-and-forget the
restart of the network service and then try to reach the SSH port. We
also need to make sure that all required packages are installed.
Finally, we drop the 'ignore_errors' parameter since we really need to
know when the networking configuration has failed.
Change-Id: I366192737b5c4e01964eaf187396eababbdc808e
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The provision-vm-nodes playbook doesn't do any VM provisioning. It
basically just bootstraps the bifrost repository so rename it to
make it clear what this is about.
Change-Id: I1c10bc6135686c5197d553a885412a94312a41bc
Signed-off-by: Markos Chandras <mchandras@suse.de>
|