refactor: when using regular nova driver, we should spin up a base solum VM and run app in docker
Bug #1334341 reported by
Paul Czarkowski
This bug affects 1 person
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| Solum |
In Progress
|
Medium
|
Unassigned | ||
Bug Description
currently when using VMs instead of nova-docker we rebuild a qcow image and inject the application into it. This is not simple and requires root on the server, and only works for the one hypervisor driver.
Instead of this we should run a base OS ( CoreOS seems like a good choice ) that is just running Docker. Docker on this box would then be told ( via cloud-init ) to reach out to the docker registry and run the appropriate docker container.
This means we'll have a consistent build process among our different nova hypervisors and will simplify things.
Revision history for this message
| Paul Czarkowski (paulcz) wrote | #1 |
| Changed in solum: | |
| assignee: | nobody → Paul Czarkowski (paulcz) |
Revision history for this message
| OpenStack Infra (hudson-openstack) wrote Fix proposed to solum (master) | #2 |
| Changed in solum: | |
| status: | New → In Progress |
Revision history for this message
| Julien Vey (vey-julien) wrote | #3 |
Revision history for this message
| OpenStack Infra (hudson-openstack) wrote Change abandoned on solum (master) | #4 |
| Changed in solum: | |
| assignee: | Paul Czarkowski (paulcz) → nobody |
| Changed in solum: | |
| importance: | Undecided → Medium |
| tags: | added: solum-deployer |
To post a comment you must log in.
Working on a MVP for this.