clusterhq/ansible-role-flocker

An ansible role to install and configure flocker

ansible-role-flocker's Issues

docker-plugin installed but inactive

after running the role:

---
- hosts: all
  vars:
    flocker_control_service_groupname: flocker_control_service
    flocker_agents_groupname: flocker_agents
    flocker_cluster_name: my_flocker_cluster
    flocker_install_docker_plugin: true
    flocker_agent_yml_path: /Users/others/ansible/project/ansible/group_vars/agent.yml 
  roles:
    - role: ClusterHQ.flocker

Installed is ok but if I check the status of the services (it is inactive)

[root@manager1 vagrant]# systemctl status flocker-container-agent
    ● flocker-container-agent.service - Flocker Container Agent
       Loaded: loaded (/usr/lib/systemd/system/flocker-container-agent.service; disabled; vendor preset: disabled)
       Active: active (running) since Fri 2016-10-14 02:57:56 BST; 760ms ago
     Main PID: 17032 (flocker-contain)
       Memory: 23.2M
       CGroup: /system.slice/flocker-container-agent.service
               └─17032 /opt/flocker/bin/python /usr/sbin/flocker-container-agent --journald

    Oct 14 02:57:56 manager1 systemd[1]: Started Flocker Container Agent.
    Oct 14 02:57:56 manager1 systemd[1]: Starting Flocker Container Agent...
    [root@manager1 vagrant]# systemctl status flocker-docker-plugin
    ● flocker-docker-plugin.service - Flocker Docker Plugin
       Loaded: loaded (/usr/lib/systemd/system/flocker-docker-plugin.service; disabled; vendor preset: disabled)
       Active: inactive (dead)

I neet to enable and restart the service manually

    [root@manager1 vagrant]# systemctl enable flocker-docker-plugin
    Created symlink from /etc/systemd/system/multi-user.target.wants/flocker-docker-plugin.service to /usr/lib/systemd/system/flocker-docker-plugin.service.

    [root@manager1 vagrant]# systemctl status flocker-docker-plugin
    ● flocker-docker-plugin.service - Flocker Docker Plugin
       Loaded: loaded (/usr/lib/systemd/system/flocker-docker-plugin.service; enabled; vendor preset: disabled)
       Active: inactive (dead)

    [root@manager1 vagrant]# systemctl restart flocker-docker-plugin

now the docker plugin is active

 [root@manager1 vagrant]# systemctl status flocker-docker-plugin
    ● flocker-docker-plugin.service - Flocker Docker Plugin
       Loaded: loaded (/usr/lib/systemd/system/flocker-docker-plugin.service; enabled; vendor preset: disabled)
       Active: active (running) since Fri 2016-10-14 02:59:40 BST; 1s ago
     Main PID: 17973 (flocker-docker-)
       Memory: 43.9M
       CGroup: /system.slice/flocker-docker-plugin.service
               └─17973 /opt/flocker/bin/python /usr/sbin/flocker-docker-plugin --journald

    Oct 14 02:59:40 manager1 systemd[1]: flocker-docker-plugin.service holdoff time over, scheduling restart.
    Oct 14 02:59:40 manager1 systemd[1]: Started Flocker Docker Plugin.
    Oct 14 02:59:40 manager1 systemd[1]: Starting Flocker Docker Plugin...

I am running CentOS Linux release 7.2.1511 (Core)

Could we add support to install from a different repository.

We have clusterhq-archive.s3.amazonaws.com hard-coded in part of this recipe. It would be sort of cool if you could override that with an environment variable or something. This would let us use these scripts to install from a development build (so we can test blog posts before releases, etc.).

make container-agent optional

By default, this role installs the flocker-container-agent by default, and if the nodes do not already have docker installed, the container agent wont start and ansible will hang. Make it optional to install the container agent especially now that flocker makes it optional to use it

add support to configure docker plugin

force installing the docker plugin is great, but we still need to upload the plugin crts and start the docker plugin by hand.