LAVA Validation Lab

Staging environment for LAVA + Celery + Fast models

Registered by Zygmunt Krynicki

We need to have a new kind of staging environment for fast models and other lava cloud-related pieces. This will include additional temporary virtual machines, physical board allocation, virtual machine allocation and proper routing for access.

Blueprint information

Status:
Complete
Approver:
Andy Doan
Priority:
Undefined
Drafter:
None
Direction:
Needs approval
Assignee:
None
Definition:
Obsolete
Series goal:
Accepted for trunk
Implementation:
Not started
Milestone target:
None
Completed by
Andy Doan

Related branches

Sprints

Whiteboard

This has been obsoleted

Unspecified topics that need consideration:

1) How to do a distributed upgrade (we should stop everything globally unless we are sure we don't need to)
2) How to run migrations without shutting everything down (launchpad disconnect trick?) List of software that cannot cope with this: copes natively, does not cope but retries safely, does not cope and blocks migration
3) How to do non-migration upgrades
4) Which changes should be done to lava-deployment-tool that would make this smoother
5) Consider adopting fabric to run production

(?)

Work Items

Work items:
Create empty oneiric "master" virtual machine: TODO
Create empty oneiric "slave" virtual machine: TODO
Setup NFS export from master that is available to slave: TODO
Install lava in the master machine: TODO
Confirm that lava continues to work in the slave machine: TODO
Put one panda in the new staging environment (permanently): TODO
Confirm that panda health job from master works: TODO
Confirm that panda health job from master, via celery to slave, works: TODO
[zyga] Configure fast model vm on master: TODO
[zyga] Confirm that rtsm (unspecified) job on master works: TODO
[zyga] Confirm that rtsm (unspecified) job on master, via celery to slave, works: TODO

Nearby

This blueprint contains Public information 
Everyone can see this information.