vm auto ha operaton when compute node broken

Registered by John Haan on 2013-08-09

Currently vm ha is progressing with manually like evacuataion function. vm-auto-ha function is auto provisioining function. For this, we should satisfy several conditions. Nova exposes a vm-auto-ha API. The monitoring tool is required. For example, ceilometer monitors the compute node. When the ceilometer detects the compute node broken, it sends a vm-auto-ha api call ( including contents like broken_host and broken_time). Nova requests All the vm in broken host to auto-ha-hosts ( extra host for vm auto ha function). For administrator, admininstrator should register auto ha host. when administrator register, auto_ha_hosts db tables is updated.

Blueprint information

Status:
Started
Approver:
Russell Bryant
Priority:
Undefined
Drafter:
None
Direction:
Needs approval
Assignee:
John Haan
Definition:
Drafting
Series goal:
None
Implementation:
Started
Milestone target:
None
Started by
John Haan on 2013-08-19

Related branches

Sprints

Whiteboard

We need to discuss this in detail on the openstack-dev list. It would also be a good candidate for a design summit session if you're able to make it there. --russellbWe need to discuss this in detail on the openstack-dev list. It would also be a good candidate for a design summit session if you're able to make it there. --russellb

Marking this blueprint as definition: Drafting. If you are still working on this, please re-submit via nova-specs. If not, please mark as obsolete, and add a quick comment to describe why. --johnthetubaguy (20th April 2014)

This looks obsolete to me, since the same problem has been addressed properly (with fencing, which is a crucial element of doing this in a way which avoids data loss/corruption) using Pacemaker rather than ceilometer: http://blog.clusterlabs.org/blog/2015/openstack-ha-compute/ --aspiers (18th Oct 2015)

(?)

Work Items

This blueprint contains Public information 
Everyone can see this information.