Merge lp:~chad.smith/charms/precise/mysql/mysql-ha-with-health into lp:~openstack-charmers/charms/precise/mysql/ha-support

[Chad Smith]

Add mysql health checks and add_to_cluster/remove_from_cluster scripts to mysql/ha-support charm. Landscape-client uses these scripts during rolling OpenStack upgrades to make sure mysql services are alive and well and that the HA vips are migrated off the current node in the event of a reboot.

Comparable changes have gone into most of the other openstack services.

[Chad Smith (chad.smith)]

We only have a very simplistic health_check in this branch because mysql service is actually completing inoperable on the passive unit due to the ceph shared rados filesystem being shifted to the active leader of the HA service. Without the filesystem, we can't validate that mysql is in good health until this unit cycles back to active leader.

[James Page (james-page)]

review: Approve

=== added directory 'scripts'
=== added file 'scripts/add_to_cluster'
--- scripts/add_to_cluster	1970-01-01 00:00:00 +0000
+++ scripts/add_to_cluster	2013-03-26 16:22:51 +0000
@@ -0,0 +1,2 @@
+#!/bin/bash
+crm node online
 
=== added directory 'scripts/health_checks.d'
=== added file 'scripts/health_checks.d/service_crm_access'
--- scripts/health_checks.d/service_crm_access	1970-01-01 00:00:00 +0000
+++ scripts/health_checks.d/service_crm_access	2013-03-26 16:22:51 +0000
@@ -0,0 +1,7 @@
+#!/bin/bash -e
+# ensure this host is present in a CRM config and has access to crm tools.
+#
+# Since mysql service isn't running on the secondary/passive node, we have no
+# other easy means for determining the health of the mysql service until this
+# node becomes the active leader for the HA service.
+crm status 2>/dev/null | grep -q `hostname`
 
=== added file 'scripts/remove_from_cluster'
--- scripts/remove_from_cluster	1970-01-01 00:00:00 +0000
+++ scripts/remove_from_cluster	2013-03-26 16:22:51 +0000
@@ -0,0 +1,2 @@
+#!/bin/bash
+crm node standby