Age | Commit message (Collapse) | Author | Files | Lines |
|
The script was refactored to don't need sudo on the
latest versions, and the documentation is still referencing it.
Change-Id: I7064317ab7b756a78638f7474f2944472d3a4ba2
|
|
We no longer run everything as root so we need to look for SSH keys
in the appropriate home directory.
Change-Id: Iae1f7eb80059e7d369c8e0c8b6c33c6a4f673f94
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Putting the host keys in '/' requires root privileges so
it's best if we place them in the same directory like the
rest of the XCI files.
Change-Id: I030ed3d6cbb57bb984a78aeffb4eca2bd5c10bb0
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
When developing XCI features it's useful to be able to use the local
repositories rather than cloning them from git since that makes
it harder to test local modifications against XCI. As such, we add
three new variables which can be used to hold local paths to the
bifrost, releng and openstack-ansible repositories. We are still
cloning the repositories but we then use the 'synchronize' Ansible
module to copy modified files from the local repositories.
Change-Id: I6d593ea48d8b9c51415d9d0848f77a498ef2f486
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The only reason we had this file was because we had to define our own
'ssh_public_key_path' variable to look in the /root directory since we
were running everything as root. Since we are now moving away from this
approach, this file is not needed anymore.
Change-Id: Ic938c4c8c877c95ce261721dc32fc5147081e236
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Since we are operating on localhost, there is no need to do that via ssh
so we use 'connection: local' instead. Moreover, we do not need to
execute everything as root so we drop the remote user directive.
Change-Id: Ib2127edad29e2da1cd1beebf42cb8f0d278ce3ad
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
There is no need to provision VM nodes as root. The only thing that we
need to run as root is the destroy-env.sh script and for that we move
it to its own play so we can use Ansible's become directive. Moreover,
since this playbook operates on localhost we can use 'connection: local'
so we don't have to execute everything via ssh and possibly speed up
the whole process.
Change-Id: Ia9efd5f30e95385b5cf193dde352f93551846c0e
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Having everything in the /opt implies root privileges. However, xci
should work as normal user and as such we use the /tmp directory as
the root directory.
Change-Id: I985ded289208e1de53ad0ba878afd5bfd745c3fd
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
In preparation for running xci-deploy as normal user, we do not need to
gain elevated privileges in order to connect to remote hosts. SSH keys
will be placed in user's ssh directory so we can ssh as normal user.
Change-Id: Ieb705515ec191955a646839e8f195d089df184bd
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
There is no good reason to run the script as root. If elevated
privileges are required then we will do it on a per case basis.
Change-Id: I7195fb3a0026ea20ce04e0b652dd3c1c56ca5741
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
This is similar to commit 6769ae9bcbf6 ("prototypes: bifrost: Add
keystone roles"). In preparation for running the bifrost jobs as normal
user we need to write a clouds.yaml file to users home directory.
This is similar to the bifrost upstream testing playbook.
Change-Id: Iae4bc73906649707667925280a19282ec5c93998
|
|
'pip install ansible' is not enough on newly installed hosts which may
lack the necessary build tools to install Ansible's dependencies. As
such, we add a script similar to the bifrost/scripts/install-deps.sh
one to pull in all the necessary distro-specific packages so Ansible
and it's dependencies can be installed from scratch.
Change-Id: I4b1e74644db9ace451ad763e4c54f1a3a43214fd
|
|
The os_client_config Ansible module used by all the ironic-* roles
depends on the keystone roles so we need to make use of them even
though they do not affect the end result. This fixes the following
OPNFV CI problem due to not having a clouds.yaml file present to
be consumed by the os_client_config Ansible module.
fatal: [controller00]: FAILED! => {"censored": "the output has been hidden due to the fact that 'no_log: true' was specified for this result"}
An exception occurred during task execution. The full traceback is:
Traceback (most recent call last):
File "/tmp/ansible_ApkCUQ/ansible_module_os_client_config.py", line 75, in <module>
main()
File "/tmp/ansible_ApkCUQ/ansible_module_os_client_config.py", line 63, in main
for cloud in config.get_all_clouds():
File "/usr/lib/python2.7/site-packages/os_client_config/config.py", line 798, in get_all_clouds
cloud, region_name=region['name']))
File "/usr/lib/python2.7/site-packages/os_client_config/config.py", line 1071, in get_one_cloud
auth_plugin = loader.load_from_options(**config['auth'])
File "/usr/lib/python2.7/site-packages/keystoneauth1/loading/base.py", line 162, in load_from_options
raise exceptions.MissingRequiredOptions(missing_required)
keystoneauth1.exceptions.auth_plugins.MissingRequiredOptions: Auth plugin requires parameters which were not given: auth_url
fatal: [opnfv]: FAILED! => {"censored": "the output has been hidden due to the fact that 'no_log: true' was specified for this result"}
Moreover, we cleanup the ~/openstack directory which may contain some
bifrost artifacts such as a clouds.yaml file.
Finally, we use 'sudo -H' because for the keystone roles we need the HOME
variable to be set properly.
Change-Id: I45b08bd33dd8ea9505fe10eb4b2b10956b3b683c
|
|
Role requirements file is used during ansible bootstrap on opnfv
host so we must ensure the right ansible role requirements is there
in advance.
Change-Id: I2c5a1edd82a51cbbe1469c31f37b5d638d32ec27
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
Pinned versions used to be used only if XCI_LOOP == 'daily', but it
seems to be required, otherwise this is very likely to fail
Change-Id: I21f638bba75846a58b12373a903d414fa89b4bc6
Signed-off-by: Juan Vidal <juan.vidal.allende@ericsson.com>
|
|
Change-Id: Iebb9945bfd9cad9ebfda127f130542f4a6334aec
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
They are exactly same now after moving nfs to compute00.
Change-Id: I9d9d2f037aa3684ba130ae0a4f9aef2e733f2d2e
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
This change
- switches cinder backend from lvm to nfs as examplified in [1].
- moves nfs shares from compute01 to compute00 in order to merge
configure-targethosts.yml playbook in upcoming change.
- updates openstack_user_config.yml to reflect the switch to nfs
and move of nfs server to compute00.
- removes exports file due to switching to lineinfile module.
[1] https://docs.openstack.org/project-deploy-guide/openstack-ansible/draft/app-config-prod.html
Change-Id: I0715c98a89ab124256e3857c1047e374f261e39b
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
For some reason, when the nodes are provisioned, they are out
of sync time-wise. This prevents neutron agents on compute nodes
from starting and then everything fails.
This change restarts chrony after the network configuration is updated
and the nodes have internet access and force synchs time on nodes.
Change-Id: Ib27b1fa0313223b52fa2e6229d5a179581d62686
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
This change
- sets static IP for controller nodes' vxlan bridges as stated
in osa documentation: compute and network hosts must have an IP [1]
- configures IPs for controller nodes' vxlan bridges.
- removes the additional vlan ip from compute nodes as it seems to
be there for all in one.
- removes glean.rules as they mess with the network configuration once
the network changes are applied.
- removes the contents of interfaces.d as osa populates config for
lxc bridges and includes this config in interfaces file thus all the
files in interfaces.d must be removed in order not to mess with the
network configuration.
- updates modules appropriately.
[1] https://docs.openstack.org/project-deploy-guide/openstack-ansible/draft/app-config-prod.html
Change-Id: I84d5f07216c8d0246da06080fb0ed19f917d6637
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
Change-Id: Ibc1b9662bb9865c0e309128327d620324ddca85c
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
Aodh is broken on the master branch due to use of conflicting
sqlAlhcemy package, causing issues during osa repo_build.
This patch disables aodh and the related services ceilometer
and gnocchi since the reason for us to have these 2 is to get
aodh.
See the failure from
https://build.opnfv.org/ci/job/xci-deploy-virtual-xenial-daily-master/26/consoleFull
Change-Id: Ic6b74328cf1692b5e26c5f30b3eb3799240c7117
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
Change-Id: If29aaf0be6ddd860450eeb10432fe75e12b1d03b
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
The bifrost, osa, and ansible role requirements sha1s are bumped
based on the HEAD of the master of the projects on April 4.
designate is also disabled based on the comment the osa commit
mentions.
https://review.openstack.org/gitweb?p=openstack/openstack-ansible.git;a=commit;h=d9e1330c7ff9d72a604b6b4f3af765f66a01b30e
Change-Id: Ib47515490c799d3b1d2b4b319277c8900cd37328
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
XCI has different jobs/loops to run
- patchset verification jobs (currently bifrost and osa in future)
- periodic jobs (bifrost and osa)
- daily jobs (for OPNFV platform deployment and testing)
The same scripts/playbooks used by XCI will also be used by developers.
We need to do different things depending on the context the scripts
and playbooks are executed.
- periodic jobs will use latest of everything to find working versions
of the components. (periodic osa will use unpinned role requirements
for example)
- daily jobs will use pinned versions in order to bring up the platform
and run OPNFV testing against it. (daily deployment will use pinned
versions and role requirements for example)
- developers might choose to use pinned versions or latest
Depending on what loop we are running, we need to do things differently
in scripts and playbooks. This variable will help us to do this in easy way.
We can of course do pattern matching of the job name but it will not
work if the scripts are used outside of Jenkins.
The default loop for non-Jenkins execution is set to daily as we want
developers to use working versions unless they change it to something
else intentionally.
Change-Id: Iff69c77ae3d9db2c14de1783ce098da9e9f0c83d
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
This patch reflects the fix Tianwei found for the keystone and gnocchi
issues with HA deployment.
https://gerrit.opnfv.org/gerrit/#/c/32367/
The keystone issue should be fixed upstream but until that happens, we
can use the patch here to move us forward.
Previously removed OpenStack services have also been enabled to
try as close deployment as possible to upstream openstack-ansible.
Change-Id: Ib7b1cfe0f363381466b43a3249667c638c09d9fd
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
This change
- fixes GIT_BASE in jobs
- adjusts build blockers
- fixes where the flavor vars should be sourced from
- sources flavor vars for bifrost periodic jobs
- renames the xcimaster node to opnfv to be consistent
- removes obsolete xci-provision.sh script
- enable additional openstack services including tempest to
ensure the deployment is sane
Change-Id: Ifdce1da52d68a26c2b87e1bd3b1996ec119d8d90
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
Change-Id: Ib5200372a44ee6a22b45c2105b3729d362eacb1d
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
Daily jobs are now switched to using xci-deploy.sh script together
with the environment variables, playbooks, and so on.
These jobs do not set any version for any of the component as
the point with these jobs is to test the platform itself, not the
tools.
DIB related stuff is still left as parameters to deploy job for SUSE
and CentOS jobs so we can override what is set by xci depending
on the distro.
Change-Id: I5a5292e6bae8fdd052858108db409e8d495fbf24
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
This change
- updates the main script to run the playbooks
- creates the initial version of the readme file
Change-Id: I4110433bce541a5d9645f34457f9e1b1ee93df19
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
This change moves preparation (cloning repos, combining opnfv/bifrost
with openstack/bifrost), destroying and creating VM nodes from the
script into a separate playbook.
This requires the host to have ansible installed. The version of ansible
to install using pip currently matches to what bifrost uses but it is
hardcoded and needs to be fixed properly.
The reason for having this as a playbook to simplify the script and
increase the reuse. This playbook will be used for
- developer sandbox
- periodic bifrost jobs to run against latest on given branch and
promoting bifrost sha1 to pin later on
- daily jobs to run using pinned versions of bifrost
Change-Id: I033f12290dfea19d4c74be80eea7203211c0369e
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
This patch adds noha target host configuration playbook, inventory, variable
files and so on to use until we have time to fix things more properly and some
other functionality becomes available in upstream. This approach will result in
duplicates but it is something we can perhaps live with.
The playbook to configure target hosts will be same with the one we will
have for noha flavor but for different for the flavor mini.
The user variables and openstack user configuration will be different between
flavors and keeping deployment specific configuration.
Change-Id: I4e2b375b9f8f6bd5f8c5da91a522b78d61a58125
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
This patch adds mini target host configuration playbook, inventory, variable
files and so on to use until we have time to fix things more properly and some
other functionality becomes available in upstream. This approach will result in
duplicates but it is something we can perhaps live with.
The user variables and openstack user configuration will be different between
flavors and keeping deployment specific configuration.
Change-Id: Ie312dfa2047d094b590de970c1436d69f4c69668
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
This patch adds noha target host configuration playbook, inventory, variable
files and so on to use until we have time to fix things more properly and some
other functionality becomes available in upstream. This approach will result in
duplicates but it is something we can perhaps live with.
NFS configuration is converted into a role and the rest is adjusted accordingly.
The playbook to configure target hosts will be same with the one we will
have for ha flavor but for different for the flavor mini.
The user variables and openstack user configuration will be different between
flavors and keeping deployment specific configuration.
Change-Id: I8faceae846285fc3e4260cbced5b740a9b0c3766
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
Incorrectly placed role configure-network causes ansible to skip
the roles at the beginning of the playbook.
Change-Id: Id85e9684eb86acad7c11104b434a319c37e16903
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
This patch adds aio opnfv host configuration playbook, inventory, variable
files and so on to use until we have time to fix things more properly and some
other functionality becomes available in upstream. This approach will result in
duplicates but it is something we can perhaps live with.
The playbook to configure opnfv host is different from the playbook used for
the rest of the flavors as our target host is the opnfv host itself. This is
the reason for not having target host configuration playbook.
This installation of aio will be driven by upstream scripts/playbooks entirely
and we just execute the scripts we have in openstack-ansible repo.
Change-Id: Ica814bdac1d324414f3add382a8241de49c93a2e
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
This effectively reverts d8235c3c97f4 and 7c1b8bf89d6ce since
upstream has now fixed the issues reported in
https://review.openstack.org/#/c/450681/
Change-Id: I35588146654d9141bc9357faa3d44e8799c5b706
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Network configuration task and accompanied handler put into a role
and handler is converted to task.
Distro specific var, interface, is introduced to ensure we do not
hardcode the interface which might not be available.
Update the templates accordingly.
Change-Id: I667620fe22c93a9b20a1d8c1b7b0051d7647b591
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
This is needed in order to configure target hosts that are in the
flavor inventory file rather than what is there by default, which
is aio.
Change-Id: Iab27ed71a3d7c03f8fbbec8ceb7af92259320412
|
|
OS family vars_files are currently empty and put there to show
the intend. (kind of TODO)
opnfv.yml holds non-distro and non-flavor specific variables.
Change-Id: I65aff2650257f2df00fd1f0a0638fd1aff596ac4
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
This patch adds the main/common playbooks, files, and templates to
be used for all flavors.
The provisioning and OpenStack installation process will be as below
- provision VMs for flavor using bifrost
- once the VMs are provisioned, configure-localhost.yml playbook will
be run, preparing the localhost in order to ensure the right playbooks
(configure-opnfvhost.yml and configure-targethosts.yml), inventory files
and var files are in place before we proceed with configuring opnfv host.
- after getting the right files for the flavor, opnfv host will be
configured using configure-opnfvhost.yml playbook.
- finally, the target hosts will be configured.
- once the above process is completed, openstack-ansible playbooks will
be run, setting up hosts, infrastructure and OpenStack.
Change-Id: I6e08b2cfdab9627f765e6fc414917b09f953cab2
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
/opt/stack is being created by bifrost so the destroy-env.sh
script should be responsible for cleaning up all the remnants.
That way we can use the destroy-env.sh script outside of Jenkins
as well.
Change-Id: I72d9737cf2446ff3ffc02f195337a6873eceb5c0
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Destroying/undefining all the VMs on host might cause issues for the
users. This change tries to reduce the possibility of it by removing
VMs listed by vbmc.
Change-Id: I5134043309722632df68a0800c6403e6f39a8d16
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
Commit d8235c3c97f4("Set database pool_size in ironic.conf and
inspector.conf templates") added a copy of the ironic.conf.j2
template file to workaround an upstream bug but it also changed
the default ironic drivers leading to broken introspection. We use
virtual BMC for controlling the VMs and this expects ipmi ironic
drivers so we fix that.
Change-Id: Ibb2cc58738ec1693e136bdce7335cd1b861ec613
Fixes: d8235c3c97f4("Set database pool_size in ironic.conf and inspector.conf templates")
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Change-Id: Ie6f079bb61905ca0458778b78fef799777b24efa
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
Reflect the changes to other impacted files as well.
Change-Id: I106f4e47fe5c75d288f4878fe6ec9f8ff39c652e
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
Updates
- RAM allocations for the flavors mini and noha increased to 12GB.
- Distro variables introduced to tool as it will be used by the CI and
the developers.
- LOG_PATH and RUN_TEMPEST are now available in user variables.
Change-Id: I4a73d7b79761d81f605cdb3fa9771ca40e9f9f02
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
Summary of changes are
- flavors directory has been removed and the flavor config files are
moved into config and renamed to <flavor>-vars
- common files are put under file
- files specific to flavors are put under file/<flavor> directories
- templates and var files are stored in template and var directories
respectively
- 3 playbooks are created
Change-Id: I8a93e0947ccb02f93a6c8f00da27e0cc6b4dc21e
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
min_pool_size and max_pool_size are undefined in ironic.conf and inspector.conf
templates, causing issues during ironic-dbsync and ironic-inspector-dbsync.
https://build.opnfv.org/ci/job/opnfv-bifrost-verify-trusty-virtual-master/224/console
This issue is there for stable/ocata as well.
These should be got rid of once the patch on openstack/bifrost is merged.
https://review.openstack.org/#/c/450681/
Change-Id: I678835e763022a7168e498047bb65d140895e019
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
This change enables node provisioning using bifrost based on the chosen
flavor.
Other changes include
- move flavor specific stuff (playbooks, inventory, vars) into their own
folders so it is easier to copy them over.
- rename flavors and get rid of xci from them.
- introduce env-vars to keep variables that are not really user variables.
Please note that this patch contains empty files and so on due to still
trying to find best way to structurei things. Apart from this, there will
probably be lots of duplications here and there which will be taken care
of once things settle.
Change-Id: I04a5d422e5b018439bafea5e68e65255ae38d22b
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|