Masakari should support creating a default failover segment
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
OpenStack Masakari Charm |
New
|
Undecided
|
Unassigned |
Bug Description
It would be quite helpful for those cloud admins new to Masakari to have a feature in the charm that would configure all compute nodes within the region as one large failure segment defaulting to the "AUTO" failover methodology.
My theory is that upon deploying Masakari with charms, it could be inferred that your compute workloads are protected by some level of default configuration of Masakari failover.
This may require both a charm configuration to set whether the charm should create such a failover segment, and what it's default settings should be, as well as an action to be run after adding new compute nodes to expand the segment host list to include new hosts.
Since failover segments in non-reserved-host modes do not define rules of VM placement during recovery, and all compute nodes under a given nova-compute charm should be able to ssh to each other for purposes of live-migration, perhaps this could come from a relation to nova-compute charms instead of masakari-charm configuration and actions.
If AZ-aware segments are necessary for a region, perhaps an option similar to ceph-mon's customized-
summary: |
- [wishlist] Configuration and action to support "Region-wide" charm- - managed failure segment + Masakari should support creating a default failover segment |