containers not started during deploy if are stopped

[Eduardo Gonzalez]

During some tests, I stopped containers with kolla-ansible stop.

Then, executed kolla-ansible deploy to start all containers. This is not the case,
containers are only started if:

- container does not exists
- check_container_differs is true, which imply removing and create new container
- config_strategy == 'COPY_ALWAYS'

    def recreate_or_restart_container(self):
        self.changed = True
        container = self.check_container()
        # get config_strategy from env
        environment = self.params.get('environment')
        config_strategy = environment.get('KOLLA_CONFIG_STRATEGY')

        if not container:
            self.start_container()
            return
        # If config_strategy is COPY_ONCE or container's parameters are
        # changed, try to start a new one.
        if config_strategy == 'COPY_ONCE' or self.check_container_differs():
            self.stop_container()
            self.remove_container()
            self.start_container()
        elif config_strategy == 'COPY_ALWAYS':
            self.restart_container()

Check_container_differs not evaluate correctly as is not noticing the container is not running.

TASK [haproxy : Check haproxy containers] ******************************************************************************************************
ok: [192.168.100.244] => (item={'key': u'haproxy', 'value': {u'group': u'haproxy', u'image': u'192.168.100.1:4000/lokolla/centos-source-haproxy:5.0.0', u'enabled': True, u'volumes': [u'/etc/kolla//haproxy/:/var/lib/kolla/config_files/:ro', u'/etc/localtime:/etc/localtime:ro', u'haproxy_socket:/var/lib/kolla/haproxy/'], u'container_name': u'haproxy', u'privileged': True}})

PLAY [Apply role kibana] ***********************************************************************************************************************

[sean mooney (sean-k-mooney)]

at least form me i hit this issue when using COPY_ALWAYS
i was able to force start eh containers by changing to COPY_ONCE then running deploy

ill then use reconfigure later to go back to COPY_ALWAYS.

can we add a dedicated kolla-ansible start/restart which will start or restart all containers in
there order they would be started in a normal deploy.

[sean mooney (sean-k-mooney)]

ok that did not work but ill get back to this on monday...

[Eduardo Gonzalez (egonzalez90)]

https://review.openstack.org/#/c/499205/

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/499205
Committed: https://git.openstack.org/cgit/openstack/kolla-ansible/commit/?id=448a10df6cbd53bc93d7b767787ef925169bc3d4
Submitter: Zuul
Branch: master

commit 448a10df6cbd53bc93d7b767787ef925169bc3d4
Author: Eduardo Gonzalez <email address hidden>
Date: Wed Aug 30 17:38:36 2017 +0200

    Add container state check in kolla_docker

    Missing container status check in recreate_or_restart_container,
    this causes if the container is not running (kolla-ansible stop),
    to not be started with deploy/reconfigure/upgrade if any other param
    changes.

    Change-Id: I5cff5f367e963ba8b1807ec46469da817e40e468
    Closes-Bug: #1714015

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/kolla-ansible 6.0.0.0rc1 release candidate.