etcd via tooz does not support group membership required by
Designate coordination.
The best k-a can do is not to configure etcd in Designate.
Change-Id: I2f64f928e730355142ac369d8868cf9f65ca357e
Closes-bug: #1872205
Related-bug: #1840070
Allow operators to use custom parameters with the ceilometer-upgrade
command. This is quite useful when using the dynamic pollster subsystem;
that sub-system provides flexibility to create and edit pollsters configs,
which affects gnocchi resource-type configurations. However, Ceilometer
uses default and hard-coded resource-type configurations; if one customizes
some of its default resource-types, he/she can get into trouble during
upgrades. Therefore, the only way to work around it is to use the
"--skip-gnocchi-resource-types" flag. This PR introduces a method for
operators to execute such customization, and many others if needed.
Depends-On: https://review.opendev.org/#/c/718190/
Change-Id: I92f0edba92c9e3707d89b3ff4033ac886b29cf6d
AArch64 needs a bit more time to boot testing instance. So give it more
time by doubling amount of connect attempts.
Change-Id: I87ca65691dfbac84349e8af24d2f36f1db1c8be1
Some services look for /etc/timezone on Debian/Ubuntu, so we should
introduce it to the containers.
In addition, added prechecks for /etc/localtime and /etc/timezone.
Closes-Bug: #1821592
Change-Id: I9fef14643d1bcc7eee9547eb87fa1fb436d8a6b3
In kolla ansible we typically configure services to communicate via IP
addresses rather than hostnames. One accidental exception to this was
live migration, which used the hostname of the destination even when
not required (i.e. TLS not being used for libvirt).
To make such hostnames work, k-a adds entries to /etc/hosts in the
bootstrap-servers command. Alternatively users may provide DNS.
One problem with using /etc/hosts is that, if a new compute host is
added to the cloud, or an IP address is changed, that will not be
reflected in the /etc/hosts file of other hosts. This would cause live
migration to the new host from an old host to fail, as the name cannot
be resolved.
The workaround for this was to update the /etc/hosts file (perhaps via
bootstrap-servers) on all hosts after adding new compute hosts. Then the
nova_libvirt container had to be restarted to pick up the change.
Similarly, if user has overridden the migration_interface, the used
hostname could point to a wrong address on which libvirt would not
listen.
This change adds the live_migration_inbound_addr option to nova.conf. If
TLS is not in use for libvirt, this will be set to the IP address of the
host on the migration network. If TLS is enabled for libvirt,
live_migration_inbound_addr will be set to migration_hostname, since
certificates will typically reference the hostname rather than the
host's IP. With libvirt TLS enabled, DNS is recommended to avoid the
/etc/hosts issue which is likely the case in production deployments.
Change-Id: I0201b46a9fbab21433a9f53685131aeb461543a8
Closes-Bug: #1729566
This patch introduces an optional backend encryption for Keystone
service. When used in conjunction with enabling TLS for service API
endpoints, network communcation will be encrypted end to end, from
client through HAProxy to the Keystone service.
Change-Id: I6351147ddaff8b2ae629179a9bc3bae2ebac9519
Partially-Implements: blueprint add-ssl-internal-network
This is a follow up to I001defc75d1f1e6caa9b1e11246abc6ce17c775b. To
maintain previous behaviour, and ensure we catch any host configuration
changes, we should perform host configuration during upgrade.
Change-Id: I79fcbf1efb02b7187406d3c3fccea6f200bcea69
Related-Bug: #1860161
Elasticsearch 6.x dropped support for mapping types[1], which by default
the Kibana index used. This means that when deploying ELK 6.x, the
Kibana index must be migrated to the new schema to preserve dashboards
and visualizations. There is a process defined[2], which involves
creating a new index with the specified schema, then reindexing the old
index's data into the new index, then doing a rename/delete.
This adds support for that workflow via Ansible. It takes place after
the ES container is restarted after an upgrade, so there will be a
(short) period of time where the Kibana index is not migrated. During
this time, Kibana still loads, but presents the user with a status
screen informing that the index needs migration.
[1]:
https://www.elastic.co/guide/en/elasticsearch/reference/6.x/removal-of-types.html
[2]: https://www.elastic.co/guide/en/kibana/6.x/migrating-6.0-index.html
Implements: blueprint elasticsearch-kibana-version-upgrade
Depends-On: https://review.opendev.org/709624
Change-Id: I4550629e2113f3da7f1cecfeab0d5fe0d899dae8
This updates the elasticsearch configuration file (and loading
mechanism) for ELK 6.x.
The default location for the configuration for all package
distributions is /etc/elasticsearch[1], so now that is where we
overwrite the elasticsearch.yml.
The path.conf and path.scripts paths are no longer supported and will
raise exceptions if utilized in 6.x.
[1]:
https://www.elastic.co/guide/en/elasticsearch/reference/6.x/settings.html#config-files-location
Implements: blueprint elasticsearch-kibana-version-upgrade
Depends-On: https://review.opendev.org/#/c/647748/
Change-Id: I4f74bfe07d4b7ca18953b11e767cf0bb94dfd67e
manila share container name variable is fixed in some places,
but in the defaults directory, manila share container_name variable
is variable. If the manila share container_name variable is changed
during deployment, it will not be assigned to container name,
but a fixed 'manila_share' name.
Change-Id: Iea23c62518add8d6820b76b16edd3221906b0ffb
The repo is Python 3 now, so update hacking to version 3.0 which
supports Python 3.
Fix problems found by updated hacking version.
Remove hacking and friends from lower-constraints, they are not needed
during installation.
Change-Id: I7ef5ac8a89e94f5da97780198619b6facc86ecfe
We don't need tox nor tell our users to use it when doing kolla
builds.
Tox is going away from base infra images.
It's already gone from aarch64 (arm64) ones.
Change-Id: I2eb5203ad93c011b8806f4b6fb56db081c14a2cb
The use of default(omit) is for module parameters, not templates. We
define a default value for openstack_cacert, so it should never be
undefined anyway.
Change-Id: Idfa73097ca168c76559dc4f3aa8bb30b7113ab28
Now that py2 is gone, oslotest dropped dependency on mock and will
soon affect Ussuri CI [1], let's use unittest.mock built in py3.
This also fixes py38 jobs and proactively prevents py36 and py37
failing due to [1]. This is because we never included mock in
test-requirements (but in lower-constraints where it does not
really belong at all) and instead relied on oslotest to bring
it in.
[1] https://review.opendev.org/716322
Change-Id: I30e82e2d87418272a71c7ee089a8acdaf8872158
Currently there are a few services that perform host configuration
tasks. This is done in config.yml. This means that these changes are
performed during 'kolla-ansible genconfig', when we might expect not to
be making any changes to the remote system.
This change separates out these host configuration tasks into a
config-host.yml file, which is included directly from deploy.yml.
One change in behaviour is that this prevents these tasks from running
during an upgrade or genconfig. This is probably what we want, but we
should be careful when any of these host configuration tasks are
changed, to ensure they are applied during an upgrade if necessary.
Change-Id: I001defc75d1f1e6caa9b1e11246abc6ce17c775b
Closes-Bug: #1860161
One way to improve the performance of Ansible is through fact caching.
Rather than gather facts in every play, we can configure Ansible to
cache them in a persistent store. An example Ansible configuration for
doing this is as follows:
[defaults]
gathering = smart
fact_caching = jsonfile
fact_caching_connection = ./facts
fact_caching_timeout = 86400
This does not affect Kolla Ansible however, since we use the setup
module which unconditionally gathers facts regardless of the state of
the cache. This gets worse with large inventories limited to a small
batch of hosts via --limit or serial, since the limited hosts must
gather facts for all others.
One way to detect whether facts exist for a host is via the
'module_setup' variable, which exists only when facts exist. This change
uses the 'module_setup' fact to determine whether facts need to be
gathered for hosts outside of the batch. For hosts in the batch, we
switch from using the setup module to gather_facts on the play, which
can use the 'smart' gathering logic.
Change-Id: I04841fb62b2e1d9e97ce4b75ce3a7349b9c74036
Partially-Implements: blueprint performance-improvements
In [1] only neutron-openvswitch-agent was fixed and not xenapi.
That merged in Ussuri and went cleanly into Train.
In Stein and Rocky, the backport was not clean and
accidentally fixed xenapi instead of the regular one.
Neither the original bug nor its incomplete fix were released,
except for Rocky. :-(
Hence this patch also removes the confusing reno instead of
adding a new one.
[1] https://review.opendev.org/713129
Change-Id: I331417c8d61ba6f180bcafa943be697418326645
Closes-bug: #1869832
Related-bug: #1867506
In stable branches we are getting hit by more py2-incompats.
Let's pin u-c in all CI pip invocations.
Change-Id: Ie2bcc7c115cd2aaf4639d90803216011b346daf3
