Age | Commit message (Collapse) | Author | Files | Lines |
|
We need to explicitly install the nfs server package instead of bundling
it to the dib image since not every node needs it.
Change-Id: I73f59f4ed582359df80ea08712b433537c087aea
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Add ability to restart the NFS server on SUSE hosts.
Change-Id: Id5f0edb399151c62302b2e36bb9451ed35501d3f
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Change-Id: I42c6f5f07ac87b5599758947fabe5fce36d44a2e
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
If OSA version is set to master, we should use the role versions
directly from upstream to ensure we do not cause issues by the
use of wrong role versions.
Change-Id: I26005d23944445bf287384f90c7f2e139224ab23
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
SUSE support is slowly being added to XCI.
Change-Id: I1ab6c01d3a33a9b2264992b5e1d202146972743f
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
In preparation for adding SUSE support, we need to configure the dib
build process to build images which match the host's OS.
Change-Id: I4944245d526f916b2033abc5d219fa7a69db54ab
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The Vagrant image may be old and have stale metadata leading to
failures when installing packages. As such, it's best to update
it as soon as possible. This is also inline with the AIO documentation
which suggests to always use an up to date OS.
Change-Id: I43d9f8431e1a6ed877936adc65c54be466b76aff
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Apparently 8G is no longer enought to deploy the 'mini'
flavor since the following error occurs:
[14505.101254] Out of memory: Kill process 23666 (qemu-system-x86) score 344 or sacrifice child
[14505.103404] Killed process 23666 (qemu-system-x86) total-vm:5566132kB, anon-rss:3008288kB, file-rss:0kB, shmem-rss:0kB
Change-Id: Iaa21ad367c63bf17a2c999082b5e09fa2e772d67
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Due to a bug in Yast2, /etc/hosts may be empty in Vagrant images so
add a workaround for it.
Change-Id: I1da9204a07f33008f9d5f1ec9ce976116e9d7a4c
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
We need access to the entire releng-xci directory so we can use the
OPNFV_RELENG_DEV_PATH variable to use modified files for testing
purposes so we move the Vagrantfile to the root directory.
Change-Id: I941b6f2efff479024ff580c3d58fcdcbadd9aa7f
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
It appears that pip install --user puts files in .local/bin
which may not always be in $PATH.
Change-Id: Ifa9fb3ee448471a00986e7f6f51544a3e436c9e1
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The xci-deploy.sh script should not be run as root.
Change-Id: I8d78d9df218508a72b6e419d16c95739fc1801e0
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
vars_files doesn't support loading files with variables in their names.
As such we add a new task to include these files before running any of
the roles. This fixes warnings like this:
Info: Starting provisining VM nodes using openstack/bifrost
-------------------------------------------------------------------------
No config file found; using defaults
skipping vars_file '../var/{{ ansible_os_family }}.yml' due to an undefined variable
skipping vars_file '../var/{{ ansible_os_family }}.yml' due to an undefined variable
skipping vars_file '../var/{{ ansible_os_family }}.yml' due to an undefined variable
skipping vars_file '../var/{{ ansible_os_family }}.yml' due to an undefined variable
skipping vars_file '../var/{{ ansible_os_family }}.yml' due to an undefined variable
skipping vars_file '../var/{{ ansible_os_family }}.yml' due to an undefined variable
Change-Id: I0a49e2ed8d811bb40a99612462752f8691133e0f
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Now that bifrost uses virtual env for Ansible we can go ahead and
bump the Ansible version to match the one from OSA so we can make use of
the latest features. Furthremore, we stop passing ANSIBLE_VERBOSITY to
the command line since it's an environment variable so Ansible already
knows about it.
Change-Id: I52d9a211fc0ced1f7830f6945b2943f35d38109b
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
bifrost forces us to use an old Ansible version which means we can't
take advantage of new features in XCI playbooks. As such, we move the
entire bifrost step to a virtual environment so we can have a newer
version of Ansible on the host.
Change-Id: I94874d600a29247b7826324cc111901995df90dc
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The command to check the galera cluster fails due to having strict
host key checking enabled.
This change disables strict host key checking for galera_container
to ensure we can verify if the database cluster is operational.
OSA does this for aio on their gate and put host key checking
configuration to group_vars which we might need to do as well.
Change-Id: Ifc59581d979c71f1db1b87126659470dc85cc2db
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
|
|
|
|
Instead of making OSA to generate self signed certs, bring our
own and pass them.
By this way we will be able to trust in that certs, and start
consuming OpenStack easily.
It will also generate proper openrc file to source it and start
consuming the cloud properly.
Change-Id: Ic72a8b05e6efb222926fc5fa0800e033b2dbd22f
Closes-Bug: RELENG-266
Signed-off-by: Yolanda Robla <yroblamo@redhat.com>
|
|
chrony randomly fails with the following issue
TASK [synchronize-time : synchronize time]
*************************************
fatal: [controller00]: FAILED! => {"changed": true, "cmd": "chronyc -a 'burst 4/4' && chronyc -a makestep", "delta": "0:00:00.004991", "end":
"2017-09-14 19:30:37.561972", "failed": true, "rc": 1, "start": "2017-09-14 19:30:37.556981", "stderr": "", "stdout": "200 OK\n503 No
such source", "stdout_lines": ["200 OK", "503 No such source"],"warnings": []}
As such, lets retry a few more times before giving up.
Change-Id: I20767bb3031061c03c9ba8dbc65859c7312bfac9
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
|
|
Using 'tee' to capture 'ssh' command output randomly breaks with the
following error:
tee: standard output: Resource temporarily unavailable
Lets just store the log on the host and copy it back in the end.
Change-Id: I41edac138b8642e8765d5c1e5974b375b5b3203b
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
|
|
|
|
Change-Id: Ie9481ce8ec891313b613d824a8f8943e706884f7
Signed-off-by: Tapio Tallgren <tapio.tallgren@nokia.com>
|
|
The SHA bump is required for SUSE support.
Change-Id: If21e7658ee2ee967c458c08d186b5c1d07b5d93c
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Update wording so it matches the upstream one and minimize the
noise on a-r-r updates.
Change-Id: I216c1ee1ac4f24ab2c5a382bfe5e2aacf324024a
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
This is a temporary fix to get noha and ha flavors working.
We should normally use bifrost inventory and have templates for
opnfv, controller, and compute groups rather than per host basis.
But since that is still pending to be done, this change tries to
make sure what we have now continues working.
Change-Id: I9132c474754f4d11d1b9e4b1288b449703994db8
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
|
|
|
|
* changes:
xci: playbooks: configure-nfs: Prepare for multi-distro support
xci: playbooks: configure-network: Prepare for multi-distro support
|
|
|
|
This change fixes the bug with cinder online migration command by moving
osa version to commit 691c59 and moving the cinder role version to
commit 96160f. [1][2]
The issue appears when the flavor is set to ha.
[1] https://review.openstack.org/gitweb?p=openstack/openstack-ansible.git;a=commit;h=691c59f25e5d271c670c24d12e8d60fa422441e1
[2] https://review.openstack.org/gitweb?p=openstack/openstack-ansible-os_cinder.git;a=commit;h=96160f084d89355ba62980b1ebbedf0eef7c7b20
Change-Id: I5339d1fd00a1b108f09c1221e5a3cbecc2f35182
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
The OSA version can be overriden by users to move to a later
version of it (such as master).
This change adds cautionary note to README to explain what
they can do in this situation.
Change-Id: I3e4943c2c4af10001b07ffd289bd300bb5f80244
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
This change moves the keepalived role version to commit b13e08 in
order to bring the fix applied in upstream, switching from iteritems
to items for python3. [1]
The flavor ha is enabled back due to incorporating the fix to xci and
the note in README.rst is removed as well.
https://github.com/evrardjp/ansible-keepalived/commit/b13e0840b09154a6d2470f71fea8eaa968525c5b
Change-Id: I6e03f579f1107475f4b4adf54699fc4f924beec3
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
download.cirros-cloud.net is having trouble and failing patchset
verification jobs.
Apart from that, we are switching to healtcheck tests using OPNFV
test framework so this patch fixes the current issue and prepares
for the future.
Change-Id: I593d3b2424250f12c8f32701007e97408e72e40d
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
There is nothing Ubuntu specific in this role apart from the nfs server
service name. As such we remove the block statement and make the service
name configurable in order to prepare for multidistribution support.
Change-Id: I3c8202315fc49b5e5afc2a0d03eaf80d83db5f4c
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Move the network templates inside the role since it's their only user.
Moreover, remove the arguments when we include the role in playbooks and
move the distro logic to the role itself.
Change-Id: I938686fdb31b9896b9e97339799a0edd4c34bf36
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Add support for testing the XCI in the latest openSUSE Leap releases.
Change-Id: I9b649b7076e031b162a10a976debf305299cf4f7
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
This change bumps osa sha to 4d39f2cc. [1] The reason for going to a later
osa sha is to include ODL and Tacker commits in upstream to ease the work
done for ODL and SFC scenarios. However, this osa sha seems to be having
issues with the flavor ha so the users should be aware of the problems
which is added into the readme as known issue and the flavor is disabled
in xci-deploy.sh.
Another reason to bump sha right now is that upstream changed a lot during
last 2 months and waiting longer will probably result in more changes, making
it harder to move to a later osa sha even harder.
Other changes included by this change are
- update ansible role requirements to move to shas pinned by upstream for pike.
- add odl and tacker roles into ansible role requirements.
- update how aio is deployed due to removal of an upstream script. (this could
perhaps have been done in a better way but the time pressure required this to
be fixed in the way as proposed.)
- ensure facts are gathered for all the hosts and containers before
running setup-infrastructure.yml playbook.
- stop grepping for failed in the log of setup-hosts.yml playbook due
to existence of rescue block in rabbitmq_cluster_create.yml [2].
- reduce the resource needs by updating the user_variables that are
based on upstream osa gate which uses aio. The values for none-aio xci
flavors will be adjusted in a separate change once the sha bump is complete.
- fix a bug in ha-vars.
[1] https://review.openstack.org/gitweb?p=openstack/openstack-ansible.git;a=commit;h=4d39f2cc29417153780210fc0bb86223387e9968
[2] https://review.openstack.org/gitweb?p=openstack/openstack-ansible-rabbitmq_server.git;a=blob;f=tasks/rabbitmq_cluster_create.yml;h=4aaad5adafa42684c0d791b49539069fe1b45635;hb=HEAD#l24
Change-Id: Ia7fc495e315e4a41359641f8be4b3c8bbf7b61fa
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
This patch
- removes obsolete openstack-ansible and puppet-infracloud directories
- adds upstream directory to keep the contributions that are pending to
be accepted by upstream in order to have progress in OPNFV. In a perfect
world, one should expect to have nothing in this directory so the items
in this folder are short-lived.
- adds prototypes directory to keep stuff that hasn't been discussed to be
part of XCI and to share ideas and trials with the rest of the community.
Change-Id: I12afe7050ff2b0ac457d4b16d21dfd7df6ac84c9
Signed-off-by: Fatih Degirmenci <fatih.degirmenci@ericsson.com>
|
|
|
|
It's possible that a previous run failed before registering the VMs
to the vbmc. As such, consulting the vbmc to retrieve the VM names will
return nothing and no VM will be deleted. As such, it's best to get
the VMs from the currently XCI profile and make sure they are really
gone.
Change-Id: If11265655f42c8a1c7627c9139ac0cbfc002b1fe
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
|
|
|
|
|
|
|
|
There is no support for openSUSE or CentOS right now yet the
script runs happily on these hosts until it breaks in horrible
ways. The only place we explicitly ask for Ubuntu hosts is the wiki
which can easily be missed by newcomers. As such, modify the script
to only run on Ubuntu hosts.
Change-Id: If9b355da51367fc95948a2b12a7f6c4d6b3bbbbf
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The default VM spec may not be good enough for every host
so allow users to override the default VM_* variables.
Change-Id: Ie846620ab81b7fedcbe4b23da735379ab75d3cc2
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
These changes are needed as part of migrating xci from releng repo to the releng-xci repo.
They cover:
- README updates
- Re-point configuration from releng/prototypes to releng-xci
The changes have been tested as follows:
- XCI_FLAVOR=aio
- OPENSTACK_OSA_VERSION=adfaa66d6108f87599e0595591b39cad2c8fb658
xci: aio has been installed
real 107m5.310s
user 11m50.180s
sys 4m11.152s
ubuntu@rack-IRA2-SymKloudBlade4:~/releng-xci/xci$
Change-Id: I8fd86c2442ee3a7c6996eedb6510dbc6eebd30b0
Signed-off-by: Dave Urschatz <dave.urschatz@cengn.ca>
|