Prototype distributed actor handles (#1137)

* Add actor handle ID to the task spec

* Local scheduler dispatches actor tasks according to a task counter per handle

* Fix python test

* Allow passing actor handles into tasks. Not completely working yet. Also this is very messy.

* Fixes, should be roughly working now.

* Refactor actor handle wrapper

* Fix __init__ tests

* Terminate actor when the original handle goes out of scope

* TODO and a couple test cases

* Make tests for unsupported cases

* Fix Python mode tests

* Linting.

* Cache actor definitions that occur before ray.init() is called.

* Fix export actor class

* Deterministically compute actor handle ID

* Fix __getattribute__

* Fix string encoding for python3

* doc

* Add comment and assertion.
This commit is contained in:
Stephanie Wang
2017-10-19 23:49:59 -07:00
committed by Robert Nishihara
parent 2f45ac9e95
commit af47737bd5
11 changed files with 799 additions and 406 deletions
+3
View File
@@ -35,6 +35,9 @@ table TaskInfo {
// Actor ID of the task. This is the actor that this task is executed on
// or NIL_ACTOR_ID if the task is just a normal task.
actor_id: string;
// The ID of the handle that was used to submit the task. This should be
// unique across handles with the same actor_id.
actor_handle_id: string;
// Number of tasks that have been submitted to this actor so far.
actor_counter: int;
// True if this task is an actor checkpoint task and false otherwise.