parallel nodes deploy
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Fuel for OpenStack |
Fix Committed
|
Wishlist
|
Fuel Python (Deprecated) |
Bug Description
When we creating new mos deploy we have dependency which told in what order we need to deploy nodes, for example:
1. Primary controller
2. Other controlers
3. Storage nodes
4. Compute nodes
I will focuses on first 2.
Deploy Primary controller take ~10min (example!!!)
Deploy Second controller take ~10min (example!!!)
Total deploy of this 2 nodes take 20min, but in each node we have several task to make, and if we finish one task usually we can do the same task to 2-nd node not wating for hole deploy.
Example of tasks order (very example)
node-1 (primary controller)
node-2 (second controller)
1. Start fuel_pkgs.pp on node-1
2. Finish fuel_pkgs.pp on node 1
3. Start globals.pp on node-1 and start fuel_pkgs.pp on node-2
3.Finish fuel_pkgs.pp on node-2
4. Finish globals.pp on node-1
5.Start logging.pp on node-1 and start globals.pp on node-2
etc.
In examples times this can reduce time need to deploy from ~20min (EXAMPLE!!) to ~12min (EXAMPLE!!)
Changed in fuel: | |
milestone: | none → 7.0 |
assignee: | nobody → Fuel Library Team (fuel-library) |
importance: | Undecided → Wishlist |
status: | New → Confirmed |
tags: | added: feature |
tags: |
added: need-bp removed: feature |
Changed in fuel: | |
assignee: | Fuel Library Team (fuel-library) → Fuel Python Team (fuel-python) |
milestone: | 7.0 → 8.0 |
status: | Won't Fix → Confirmed |
no longer affects: | fuel/8.0.x |
tags: | added: area-python |
This issue is related to orchestrator and how we call granular tasks. However, I'd say that some parallelism is implemented as we deploy primary-controller and then all other controllers in parallel.