* Add sunbeam project template to run pep8, py3 tests
* Add zuul.d/zuul.yaml to run pep8, py3, cover tests
* Update charmcraft and requirements for each charm
* Add global tox.ini to invoke fmt, pep8, py3, cover,
build
* Add gitreview file
* Fix py3 test failures in ciner-ceph-k8s, glance-k8s,
openstack-exporter
* Add jobs for charm builds using files option so that
job is invoked if files within the component are
modified. Add charm builds to both check and gate
pipeline.
* Make function tests as part of global. Split the function
tests into core, ceph, caas, misc mainly to accomodate
function tests to run on 8GB. Add function tests as
part of check pipeline.
* Add zuul job to publish charms in promote pipeline
Add charmhub token as secret that can be used to
publish charms.
Note: Charmhub token is generated with ttl of 90 days.
* Run tox formatting
* Make .gitignore, .jujuignore, .stestr.conf global and
remove the files from all charms.
* Make libs and templates global. Split libs to internal
and external so that internal libs can adhere to
sunbeam formatting styles.
* Add script to copy common files necessary libs, config
templates, stestr conf, jujuignore during py3 tests
and charm builds.
* Tests for keystone-ldap-k8s are commented due to
intermittent bug LP#2045206
Change-Id: I804ca64182c109d16bd820ac00f129aa6dcf4496
Currently the heat charm instance runs heat-api and
heat-engine services or heat-api-cfn or heat-engine
services. Change heat charm to deploy heat-api,
heat-api-cfn, heat-engine containers.
Change the ingress relation to use traefik-route
interface instead of ingress interface so that the
traefik configuration for heat-api and heat-api-cfn
serive can be written by charm.
Add heat-api-cfn pebble container handler and update
service endpoints accordingly.
Remove heat-config interface and corresponding handlers.
Change-Id: I391f8d4ffefcebdb2423fcc1947590ca906d711a
Stop all the services running in the container
when mandatory integrations are removed.
Currently the relation data is not cleared during
processing of gone away events. This is a bug in
juju reported at [1]. Due to this the relation is
still considered as ready.
Add workaround to check the event and mark the
relation to not ready if the event is gone away
event of the relation.
Catch database relation-broken event and block the
application.
[1] https://bugs.launchpad.net/juju/+bug/2024583
Change-Id: I80a0120d08b79561c13996c7a0f055824a1d5336
Implement stop-services function for the charm.
For relations amqp, ovsdb-cms, reset the
corresponding snap data.
Add mandatory_relations to openstack-hypervisor
charm.
Depends-On: https://review.opendev.org/c/openstack/charm-ops-sunbeam/+/899012
Change-Id: Ie6a735d4cad281c080e47208cf87f34e50d00fd1
Recent change Ieffac3732c95846d0e3ec2408a9be4d96e477603 broke the
neutron-k8s py3 tests due to self.private_key being
accessed (_request_certificates method, `if self.private_key`)
before it had been properly set causing an attempt to decode None
Change-Id: I33bbf65e1d170cbc80161c573a381db43468b413