Goals:

1. General infrastructure
2. Fix oncall schedule (@pcarranza)
3. Mountpoint with CephFS in production.
4. Some project stored in CephFS in production.
5. Initial prometheus with at least blackbox testing covered in chef.
6. Get backups under control - at least having a path forward and a full understanding.
7. Integration testing
8. Performance
9. MR Diff latency under 7s
10. MR commits latency under 5s
11. Improve performance of post received task
12. ssh access stable timing (in the new NFS mountpoint)

Risks:

1. We don't have a clean metric for the post received task, so the goal is incomplete.
2. We don't have a sane metric to expect out of the ssh access timing, so I'm good with not having a range from 2 seconds to 26, but make it stable at some point in the middle consistently. We added blackbox monitoring to get some sense out of it.
3. Integration testing is badly defined and is not actionable, we need to better define the goal here.