Age | Commit message (Collapse) | Author | Files | Lines |
|
Deployers may want to use a different DNS server so allow them to
override the ipv4_nameserver option. If the variable is not set,
then we use the libvirt DNS if we are behind a proxy, otherwise
we default to the Google DNS.
Change-Id: I96cf63758902d4aae3d155b2e8beef650449ebc9
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The old role has disappeared so we have to use another one.
Change-Id: I8533912a19680e93a9cf061109019a0801485936
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
This change renames functest prepare-tole to prepare tests and makes other
adjustments to the role to move common test preparation steps to its own
script so we can prepare for Functest and Yardstick at one go.
Similar things are required to be prepared for running Functest and Yardstick
such as
- installed packages
- external network creation
- creation of run-functest.sh and run-yardstick scripts from templates
- preparation of environment variables
This change will fail verification until the changes below is submitted.
https://gerrit.opnfv.org/gerrit/#/c/61645/
Change-Id: Id1020d3e61abd3f087863c06a132c5021339d655
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
|
|
This reverts commit 42501f0ef7e0f0729b1c780102fb9713ef383fb3.
This also removes the entire SSL management code and we let the
haproxy_server role generate the certificates for us.
We also need to bump the openrc role to include an upstream patch
which fixes the openrc template file.
deploy-scenario:os-nosdn-nofeature
installer-type:osa
Change-Id: I9bb590c9f1d5bc63519cfb4794dc15f794cc5b07
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The bifrost and ironic SHAs are from the Stein release which is still
in early days of development. We should use Rocky SHAs for the time
being. However, bifrost from Rocky does not support Ansible > 2.4 so
we have to use bifrost from master branch but all the ironic components
have to be from the Rocky release.
Change-Id: I1acb28e4a292c774d681519517be19a2da230851
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
|
|
|
|
Functest download_images.sh script downloads images that are
not needed for functest-smoke so we only download the necessary
images to cut the time down.
deploy-scenario:os-nosdn-nofeature
installer-type:osa
Change-Id: I0be643c4ccd4b8009e68433f5d635231afd2550a
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
This change
- bumps kubespray version to 8b3ce6e418ccf48171eb5b3888ee1af84f8d71ba, taking
Kubernetes version from 1.10.2 to 1.10.4.
- pins Ansible version used by Kubespray to same Ansible version as the rest.
The var kept as is in case if we need to pin Kubespray to a different Ansible
version in future.
installer-type:kubespray
deploy-scenario:k8-calico-nofeature
Change-Id: Iefbe82f0b9b10296e4362bca07630047c56630ac
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
|
|
This change
- bumps OSA SHA to b9d9269528ecbe04b2638a73c2b0f49839f27422
- bumps bifrost SHA to 0d0cfd908d10a670668619c575fd338d1ff328b7
- Pins Ansible to 2.5.8. OSA is using 2.5.5 which has a bug in the pause
module so we need to use something newer.
Change-Id: Idf05bbef880db11de4a41464ae5080aa21a0613c
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
|
|
The role used to get the output of 'virsh list --all' to determine how
many VMs are present and shut off in the system. This takes *all* VMs
on the system into consideration so it may skip creation of some or all
of the XCI VMs if we happen to have other VMs present. We can improve
the situation by simply dropping this check and always provision the
VMs we want. If the VM is already present, then the module will simply
do a sanity check of its configuration. This allows XCI to run alongside
other VMs.
Change-Id: I54255a1959509671c0305f48f23a55b6e900684f
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
|
|
According to OpenStack admin is the network for pxe boot and mgmt is the
network for OpenStack services to communicate. We were using both in XCI
indistinctly
Change-Id: I3959e767098ac2be7161a5e84735fde9ab129784
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
|
|
using kubespary to deploy the istio, kubespary support to deploy istio
installer-type:kubespray
deploy-scenario:k8-nosdn-istio
Change-Id: Id8c04936187c89fafa921dada382a0e9e11aab27
Signed-off-by: wutianwei <wutianwei1@huawei.com>
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
The info is logged to the console before post-deployment tasks and
this makes it seeing the deployment status and login info inconvenient.
This change moves the logging to the end, after post-deployment.
Change-Id: Ic7e232a0b8343e666f54087fbcc09481ab59ca90
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
This change removes OSM variables from releng-xci repo since they
are declared in os-nosdn-osm scenario itself in releng-xci-scenarios
repo.
In order for this change to be verified, below change must be submitted.
https://gerrit.opnfv.org/gerrit/#/c/61205/
installer-type:osa
deploy-scenario:os-nosdn-osm
Change-Id: I9f3a1dbdfecfb2bb08cc215758bdc52aefcdf697
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Functest Smoke test requires additional images to be available
during testing. One of the images is decompressed using xz and we
need this to be available on deployment host in order to be able
to have the images available for Functest execution.
Change-Id: I5647b3bef37fc55e8c5cc9aec5d0b2c3ea628b8a
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change our idf to be more aligned with lf's idf. Adapt
dynamic_inventory.py to the change
Change-Id: Ib8f6d1684a00a8eb5ae06d5d04d308d4325cd444
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
This change
- bumps OSA SHA to cbfdb7dc295ff702044b807336fab067d84a3f20
(mostly based on Rocky RC1)
- bumps bifrost SHA to c1c6fb7487d5b967624400623fd35aabf303b917
- pins Ansible to 2.4.6.0
- switches to ollivier/functest-healtcheck since OS is bumped to Rocky
Change-Id: Icc14e3e794b489dafd78b426c54051a3732ccb1a
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Deployment not working because of a bug. This bug gets fixed with the
new SHA
Change-Id: I91c6df6bf7ea10492265466ebdfbf4264492936b
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
Version 1.4 brings in additional dependencies and it's currently
not used in OpenStack anyway. So lets stick to 1.3 for now.
Change-Id: I2489168cae12f7fa3271c2de7d4fcf37bdb97810
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Change-Id: If87afa892a0c39d082cb1563e1c019f46b742169
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
* changes:
xci: xci-destroy-env.sh: Update virtualbmc path
xci: create-vm-nodes: Install virtualbmc in the XCI virtualenv
xci: osa: Drop openSUSE mirror variables
|
|
|
|
|
|
Bifrost no longer installs on the deployment host so we don't need
to look into the bifrost venv for a vbmc installation. Moreover, XCI
uses virtualenv by default, so we should only look there for vbmc.
Finally, we should clear vbmc config files from previous deployments
since that could leave artifacts behind.
Change-Id: I4cd35636a65e1823c9e4ca9a7c5b17d4b103c6a2
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
XCI prepares a virtualenv for us, so we should install virtualbmc
in it.
Change-Id: I320d1c7cad9c5c821269b55252cb7ab4f5136f40
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
we let downloads.opensuse.org redirect us to a good one since
the hardcoded one does not appear to work anymore. Ansible gets
upset if we feed it an empty variables file, so we set a couple
of variables there to make it happy.
Change-Id: I887522ebc71dd866d544e75beeff47af6111e059
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Leap 15 is using ID=opensuse-leap and Tumbleweed is using
ID=opensuse-tumbleweed so we should extend the regexp to match
these distributions as well. This allows XCI to support more
openSUSE variants as deployment hosts.
Change-Id: I8bf8e7b15fc00c45807ff92fbc14d7ec79ec4bcb
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
This change integrates os-nosdn-osm scenario.
The installation of OSM is done as post-deployment. This is achieved by
the addition of the new playbook named post-deployment.yml in scenario
role folder.
This mechanism is available for all OpenStack scenarios. If anything
needs to be done for a specific scenario as part of the post-deployment,
it can be achieved by creating playbook post-deployment.yml.
If post-deployment.yml exists in scenario role, the framework will run
it once the regular deployment playbooks are run successfully. If the
file does not exist, it will be skipped.
The location of the post-deployment.yml is
releng-xci-scenarios/<scenario>/role/<scenario>/tasks/post-deployment.yml
This is only implemented for OpenStack scenarios currently and similar
mechanism is needed for K8S scenarios as well which will be implemented
in a separate change.
installer-type:osa
deploy-scenario:os-nosdn-osm
Change-Id: I16780abffca39699eb2f38f662479f0e4d551504
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
|
|
Upgrading the host could bring undesired side effects so we should
not do that as part of CI deployment.
Change-Id: Ib776cdf96943e22155c2dd47fdadb3e8a86f6143
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
OPNFV VM creation is separated from the rest of the nodes, resulting
in its disk being left without resizing. It is important to resize
its disk to the value defined in its PDF opnfv_vm.yml so the installation
of the other tools does not fail due to lack of space.
Change-Id: I8300e6e355d11788cc983fcebca56076e89918e1
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
|
|
|
|
The self-signed certificates are causing troubles so until we implement
a proper certificate chain in XCI we should disable the SSL endpoints.
Change-Id: Ife4ef78de1569121c435a806924a3f90917fd85b
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The flavor all-in-one (aio) hasn't been maintained well and
the stability of it is in question. This change disables the
flavor aio and sets the default flavor to mini.
It is important for XCI to enable the aio back in order to help
users who want to try XCI but don't have big enough machine to
go for bigger flavors. However, we need to verify all flavors
for acceptable user experience so the flavor can only be enabled
back once this happens.
Change-Id: I053c8fec78876dea71439f1f1bd737a105254ad4
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
This patch is doing the first work item of the spec:
https://github.com/opnfv/releng-xci/blob/master/docs/specs/infra_manager.rst
It creates the required VMs by XCI to afterwards deploy the VIM. It does that
by reading the pdf provided by the user.
- It is currently assumed that the OS for the VM will be installed in the first
disk of the node described by the pdf
- It is assumed that the opnfv VM characteristics are not described in the pdf
but in a similar document called opnfv_vm.yml
- All references to csv from bifrost-create-vm-nodes were removed
Change-Id: I46a85284e4ce7df21cbf66f66619b35f74251e68
Signed-off-by: Manuel Buil <mbuil@suse.com>
Co-Authored-by: Markos Chandras <mchandras@suse.de>
|
|
We may need to perform non-root tasks on the nodes, so we should have a
regular user present as well.
Change-Id: I0b824f1875ce04fe67424fc96e754e06da53be8d
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The OPNFV VM acts as a sandbox for installing and running bifrost so
it is pdf agnostic. As such, we use its own descriptor file to provision
this node.
Change-Id: Ice33dc744c04147f9c746285c583de56b8353cc3
Co-Authored-by: Markos Chandras <mchandras@suse.de>
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
In preparation for using PDF/IDF to deploy nodes in XCI, we need to have
shade and python-pyyaml on the host since these are requirements for the
OpenStack Ansible modules (eg os_ironic) and the dynamic inventories.
Change-Id: Ie26af0dc58599af6e29a47d91707976dcf16ac6d
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Add a new role based on the bifrost one to create nodes for the bifrost
virtual deployments. This role will install and configure libvirt on the
host, download a prebuilt OPNFV VM image and deploy the OPNFV VM using
that image. Moreover, it will create the rest of the nodes for the
virtual deployment which will be configured by bifrost later on.
Change-Id: I9fbd084261351d3b53ae373060f43df046191c5e
Co-Authored-by: Markos Chandras <mchandras@suse.de>
Signed-off-by: Manuel Buil <mbuil@suse.com>
|
|
Functest changes have significant impact which blocks everything in XCI so this
change pins the image to a known sha to get the original set of healthcheck
testcases until the impacts are analysed and concerns are raised to Functest
and the wider OPNFV Community and addressed based on community consensus.
Pinned version of functest-healthcheck contains the test cases below.
- connection_check
- api_check
- snaps_health_check
deploy-scenario:os-nosdn-nofeature
installer-type:osa
Change-Id: Ic9222af8c27e58491b7b60a7504df9d792b5e753
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
|
|
In order to save us some time and bandwidth we can copy the deployment
image to the clean VM in order to be used by bifrost to prepare the
virtual machines. Moreover, we also put our key there so we can ssh
as the root user.
Change-Id: Id21e19aaefde71862c21180129c664c1bf121c32
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
|