Convergence: optimize RPC calls to worker
Bug #1501192 reported by
Anant Patil
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
OpenStack Heat |
Triaged
|
Wishlist
|
Anant Patil |
Bug Description
RPC calls to worker to work on a resource should be optimized. Currently, all the check_resource calls do a RPC to worker. This can be optimized:
- When only one engine-worker is configured, the worker calls can be local instead of RPC. This saves lot of round-trips to AMQP.
- When a worker can steal a lock for a resource, it should right away run the task instead of doing another RPC call.
- If all the children resources are done, the current worker should pick-up the parent and start. If the current resource satisfies multiple parents, the current worker should take one and distribute remaining.
summary: |
- Comvergence: optimize RPC calls to worker + Convergence: optimize RPC calls to worker |
Changed in heat: | |
assignee: | nobody → Anant Patil (ananta) |
Changed in heat: | |
milestone: | mitaka-1 → mitaka-2 |
Changed in heat: | |
milestone: | mitaka-2 → mitaka-3 |
Changed in heat: | |
milestone: | mitaka-3 → mitaka-rc1 |
Changed in heat: | |
milestone: | mitaka-rc1 → newton-1 |
Changed in heat: | |
milestone: | newton-1 → newton-2 |
Changed in heat: | |
milestone: | newton-2 → ongoing |
Changed in heat: | |
importance: | Medium → Wishlist |
To post a comment you must log in.
Anant, this bug description looks like for several related bugs. May be will be better to add spec about it and upload three different patches?