Rework data transmission logic during action execution

Registered by Yanyan Hu

This blueprint plans to rework the data transmission logic during action running progress including pre/post /execute. The basic principle is:

1. All data that generated BEFORE action execution by policies' pre_op or from rpc request should be stored in action.input and will be consumed by action execution;
2. All data that generated DURING action execution and will be consumed by action routine itself should be stored in action.data;
3. All data that generated DURING action execution and will be consumed by policies' post_op should be stored in action.output.
4. Policy's pre_op should only consume data stored in action.input while post_op should only consume data stored in action.output;

Blueprint information

Status:
Complete
Approver:
None
Priority:
Undefined
Drafter:
Yanyan Hu
Direction:
Needs approval
Assignee:
Yanyan Hu
Definition:
Obsolete
Series goal:
None
Implementation:
Unknown
Milestone target:
None
Completed by
Yanyan Hu

Related branches

Sprints

Whiteboard

This is just a bug of LB policy and we just need a bug fix for it.
https://bugs.launchpad.net/senlin/+bug/1535508

The design is like this: both action.inputs and action.outputs are for enduser information. User can get them by getting action detail. However, action.data is invisible for enduser. It is used to transmit data between multiple policies and action execution routine.

(?)

Work Items

This blueprint contains Public information 
Everyone can see this information.

Subscribers

No subscribers.