[RFE] expose a configuration option to set pacemaker in maintenance mode

[Felipe Reyes]

In a long running environment maintenance is part of its lifecycle, for example when firmware updates in switches is needed. So the ability to set pacemaker in maintenance mode at the cluster level[0] and at the unit level[1]

I think the global cluster maintenance status should be managed via a configuration option (juju config hacluster maintenance-mode=True) while the unit level through an action (juju run-action hacluster/0 maintenance-mode --enable|--disable).

In both cases the unit's status message should reflect the internal state of pacemaker.

[Other Info]

* https://www.hastexo.com/resources/hints-and-kinks/maintenance-active-pacemaker-clusters/

[0] crm configure property maintenance-mode=true
[1] crm node maintenance juju-xxx

[Felipe Reyes (freyes)]

Why is the global maintenance mode (cluster level) managed through a config option while the node level through an action?

 - Application level manipulation feels more appropriate in a config flag that will be able to express the current state of the application, but actions are executed against a unit.

I believe there is not "wrong" approach on this, this is more a thing of design decision.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/476151

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/476151
Committed: https://git.openstack.org/cgit/openstack/charm-hacluster/commit/?id=e95488afa0c8985ec494b1c53fe80ed2e6bbd728
Submitter: Jenkins
Branch: master

commit e95488afa0c8985ec494b1c53fe80ed2e6bbd728
Author: Felipe Reyes <email address hidden>
Date: Wed Jun 21 09:29:21 2017 -0400

    Add maintenance-mode configuration option

    This config option allows syadmins to set pacemaker in maintenance mode
    which will stop monitoring on the configured resources, so services
    can be stopped/restarted and pacemaker won't start them again or
    migrating resources (e.g. virtual IPs).

    Change-Id: I232a043e6d9d45f2cf833d4f7c4d89b079f258bb
    Partial-Bug: 1698926