The proposed approach allows for checking whether config
files are current, e.g. cases when the deployment was aborted after
config files were generated but before they were injected into the
containers which lead to old config staying in containers.
After this patch we can do:
kolla-ansible genconfig
kolla-ansible deploy-containers
and it would do what we expected rather than being a noop
in the second part.
We also lose the need to have notifies
and whens in config and handler sections respectively.
This is optimised in a separate patch.
Future work:
- optimise for large files
- could we get away with comparing timestamps and sizes?
container's should have a newer timestamp due to copy,
could also preserve it
Change-Id: I1d26e48e1958f13b854d8afded4bfba5021a2dec
Closes-Bug: #1848775
Depends-On: https://review.opendev.org/c/openstack/kolla/+/773257
Co-Authored-By: Mark Goddard <mark@stackhpc.com>
It will allow us to fail fast when pulling the image
is a problem - instead of failing in the middle of
deployment.
Change-Id: I017cddcfbbc5449e63d807385216b94e74503c9b
Adds HAcluster Ansible role. This role contains High Availability
clustering solution composed of Corosync, Pacemaker and Pacemaker Remote.
HAcluster is added as a helper role for Masakari which requires it for
its host monitoring, allowing to provide HA to instances on a failed
compute host.
Kolla hacluster images merged in [1].
[1] https://review.opendev.org/#/c/668765/
Change-Id: I91e5c1840ace8f567daf462c4eb3ec1f0c503823
Implements: blueprint ansible-pacemaker-support
Co-Authored-By: Radosław Piliszek <radoslaw.piliszek@gmail.com>
Co-Authored-By: Mark Goddard <mark@stackhpc.com>
Often cephadm jobs fail with:
Mar 30 13:01:21 primary bash[75459]: debug 2021-03-30T13:01:21.844+0000 7fa30431f700 -1 error: monitor data filesystem reached concerning levels of available storage space (available: 4% 1.8 GiB)
Let's check if 5G OSD helps and also print df -h output for reference
Change-Id: I6960fd0f378aea5a14a73d9228edf86fb86cac6c
For using 3rd party Octavia providers (such as OVN provider) an
octavia-driver-agent container must be running to expose those providers to
use.
OVN CI job has been extended with deploying Octavia and testing OVN Load
Balancer.
Closes-Bug: #1903506
Depends-On: https://review.opendev.org/c/openstack/kolla/+/771191
Change-Id: Ibafa8b7307981f2a51e630cc113d18af6162171c
Now that the issue is fixed upstream, let's remove the workaround.
[1] If3943060b5d09bd153b6401d34c7d10d3dc864fe
Change-Id: I9cbeee5a397d736338ff2065001d9d6be20cb66e
ELK 7 requires some minor changes from the existing ELK 6 config.
Depends-On: Icfa3db5788b25f70ee75411dbaf20d8d4a6a734b
Change-Id: I9815d202a77da0477aea43d714a5def8a24724fa
- Remove /var/log/kolla link to omit uploading the same logs twice
- Remove ARA HTML report (usually takes around 120MB) - can be easily
generated from the sqlite db
Change-Id: I74cd6d1128689ab2c73f00ee08af3778d7d670a4
In order to disable libvirt debug in CI (which takes vast amount of storage)
this change introduces nova_libvirt_logging_debug and disables that in CI.
Change-Id: I90bfd1b300ad3202ea4d139fda6d6beb44c5820f
The merge_configs action plugin is updated with an option to control
whether whitespace should be used around equal signs. The default
remains to use whitespace: only zookeeper.cfg doesn't use them.
This is to avoid issues with the zkCleanup.sh script which expects no
whitespace. The ZooKeeper documentation also uses no whitespace in
configuration file examples.
Change-Id: Ia082a1c002cc4e8b04f7696fdee827b747c6d13f
Closes-Bug: #1917490
This reverts commit ff441c1c0ceb8fc06a17f5db8e460ad00e8db365.
Since RabbitMQ TLS is still not functional in Victoria, it is
not enabled for upgrade jobs.
Change-Id: I575942c8d90441145de78dcb16a2b4c1f172773b
Installing ceph-ansible in the virtualenv on CentOS 8 fails with:
ModuleNotFoundError: No module named 'setuptools_rust'
This error appeared following the release of cryptography 3.4, which now
includes Rust code. It can be installed without Rust using a Python
wheel, but only with more recent pip than version 9.0.3 available as RPM
on CentOS 8. The cryptography bug report [1] recommends pip>=19.1.1.
This change upgrades pip in the virtualenv before installing
ceph-ansible.
[1] https://github.com/pyca/cryptography/issues/5753
Change-Id: I47473de6f71c422db2238d653c2d8f379c55e79b
Installing kolla-ansible system-wide on CentOS 8 fails with:
ModuleNotFoundError: No module named 'setuptools_rust'
This error appeared following the release of cryptography 3.4, which now
includes Rust code. It can be installed without Rust using a Python
wheel, but only with more recent pip than version 9.0.3 available as RPM
on CentOS 8. The cryptography bug report [1] recommends pip>=19.1.1.
This change switches to using pip --user when installing kolla-ansible.
Also fixes an issue with ansible-lint which was failing on
etc/kolla/globals.yml due to a missing space before comments.
[1] https://github.com/pyca/cryptography/issues/5753
Change-Id: Ifaf1948ed5d42eebaa62d7bad375bbfc12b134d5
Closes-Bug: #1915141
Adds the following new Zuul job for testing deployment of Monasca and
associated services:
* kolla-ansible-centos8-source-monasca
All core OpenStack services except for Keystone are disabled to ensure
enough memory is available.
A follow up patch will replace the basic tests here with Tempest.
Co-Authored-By: Doug Szumski <doug@stackhpc.com>
Change-Id: I5d33fd3d7b69798ba0aa23509f7b809065f61c19
To gain visibility into how our upgrades affect existing Cinder
volumes, a new testing path is required.
This patch adds it.
Additionally, it refactors the repeated actions and ensures that
we wait for volume deletions as well.
Change-Id: Ic08d461e6fdf91c378a87860765a489c2f86d690
Related-Bug: #1904062
