store etcd data in a persistent volume
Bug #1697655 reported by
Ricardo Rocha
This bug affects 2 people
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| Magnum |
New
|
Undecided
|
Ricardo Rocha | ||
Bug Description
In principle a multi instance etcd service does not require persistency for its storage.
But on a cluster with a single master, no LB setup, allowing the etcd storage to be put in a persistent volume opens up the possibility of upgrading the master (including the instance image) by doing a simple nova rebuild. Both the IP and the etcd data would be preserved, with minimum downtime for the cluster APIs.
To cover this use case, we should add an option to the cluster creation where users can specify the size of the volume for the etcd storage. Default should stay 0, which means no persistent volume should be created and the local instance disk is used as happens today.
| Changed in magnum: | |
| assignee: | nobody → Ricardo Rocha (rocha-porto) |
Revision history for this message
| OpenStack Infra (hudson-openstack) wrote Related fix proposed to magnum (master) | #1 |
Revision history for this message
| OpenStack Infra (hudson-openstack) wrote | #2 |
Revision history for this message
| OpenStack Infra (hudson-openstack) wrote Related fix merged to magnum (master) | #3 |
Revision history for this message
| OpenStack Infra (hudson-openstack) wrote | #4 |
To post a comment you must log in.
Related fix proposed to branch: master
Review: https:/