Commit graph

105 commits

Author SHA1 Message Date
Robert Nishihara
0a44145906 Fix the resetting of reusable variables on the driver and cache functions to run on all workers. (#446)
* Properly reset reusable variables on the driver when remote functions are run locally on the driver.

* Cache functions to run on all workers that occur before ray.init is called.
2016-10-12 22:17:22 -07:00
Robert Nishihara
292656013a Suppress exceptions in the error logging thread when program exits. (#432) 2016-09-15 13:48:23 -07:00
Wapaul1
d5815673a5 Changed ray.select() to ray.wait() and its functionality (#426)
* Re-implemented select, changed name to wait

* Changed tests for select to tests for wait

* Updated the hyperopt example to match wait

* Small fixes and improve example readme.

* Make tests pass.
2016-09-14 17:14:11 -07:00
Robert Nishihara
ba56b08474 Reintroduce passing arguments by value to remote functions. (#425)
* Reintroduce passing arguments by value to remote functions.

* Check size of arguments passed by value.

* Fix computation graph visualization.
2016-09-10 21:11:18 -07:00
Robert Nishihara
5802cab87c Fix bug in which ObjectFixture gets called at exit after raylib gets set to None. (#416) 2016-09-07 18:49:19 -07:00
Robert Nishihara
d264713ceb Work around Arrow bug by increasing metadata size. (#415) 2016-09-07 18:46:04 -07:00
Robert Nishihara
11a8914684 Allow users to serialize custom classes. (#393)
* Allow serialization of custom classes.

* Add documentation and test cases, also fix pickle case.

* Don't allow old-style classes.
2016-09-06 13:28:24 -07:00
Robert Nishihara
d5cb3ac090 Propagate error messages from functions that run on all workers. (#410) 2016-09-06 10:06:43 -07:00
Robert Nishihara
327d7ff689 Fix bug to enable calling ray.get multiple times on same ObjectID. (#409) 2016-09-04 13:32:55 -07:00
Philipp Moritz
68cec55a98 Refcount without modifying objects (#407)
* refcount without modifying objects

* add documentation

* Update tests and documentation.

* Remove extraneous code.

* Update numbuf version.
2016-09-04 12:07:52 -07:00
Robert Nishihara
81f40774a7 Remove ObjectID aliasing from the API. (#406)
* Remove ObjectID aliasing from the API.

* Update documentation to remove aliasing.
2016-09-03 19:34:45 -07:00
Richard Shin
efb61ca9c7 Return scheduler address in ray.init (#403) 2016-09-03 18:25:47 -07:00
Richard Shin
80fdfcd1a5 Check num_objstores > 0 in start_ray_local (#402) 2016-09-03 17:37:06 -07:00
Philipp Moritz
3548797202 [API] Implement get for multiple objects (#398)
* [API] Implement get for multiple objects

* Small fixes.
2016-09-02 18:02:44 -07:00
Robert Nishihara
fb7ccef493 Allow remote decorator to be used with no parentheses. 2016-08-30 16:38:26 -07:00
Robert Nishihara
b87912cb2f Remove typing module. 2016-08-29 22:16:19 -07:00
Robert Nishihara
d7f313a026 Remove type information from remote decorator. 2016-08-29 22:05:59 -07:00
Wapaul1
420bcc0477 Remote function returning non-serializable type no longer shuts worker down (#384)
* Moved put_objects in main_loop to inside of try block

* Added test for failed serialization

* Fixed naming

* Minor
2016-08-25 15:26:22 -07:00
Robert Nishihara
b1db890cd8 Export GRPC environment variable in Python instead of in setup-env.sh (#382)
* Export GRPC logging environment variable in Python.

* Don't wipe PYTHONPATH in setup-env.sh
2016-08-18 15:39:19 -07:00
Robert Nishihara
e06311d415 Automatically add relevant directories to Python paths of workers (#380)
* Make ray.init set python paths of workers.

* Decouple starting cluster from copying user source code

* also add current directory to path

* Add comments about deallocation.

* Add test for new code path.
2016-08-16 14:53:55 -07:00
Wapaul1
7246013008 Implement select to enable waiting for a specific number of remote objects to be ready. (#369) 2016-08-15 16:51:59 -07:00
Robert Nishihara
b29fc0c481 Use random string for worker c++ logfile. (#378) 2016-08-15 15:55:34 -07:00
Robert Nishihara
87bb7a8f67 [WIP] Large changes to make the tests pass. (#376)
* Revert "Make tests more informative (#372)"

This reverts commit fd353250c8.

* fix bugs, in particular deactivate worker service on driver and remove condition variables

* changes to minimize the changes in this PR

* switch from faulty mutex synchronization to using atomics

* Increase the default size of the message queues, to accommodate exporting large numbers of remote functions. This is a temporary fix, but not a long term solution.

* Reorganize the scheduler export code to queue up exports. This does not solve the underlying problem yet, but sets up a solution.

* Start a separate thread on driver to print error messages by constantly querying the scheduler. This is a temporary solution because the solution based on starting a worker service for the driver which the scheduler can push error messages to is buggy.

* Fix segfault in taskcapsule destructor.

* Move tests for catching errors into a separate test file.

* Revert "roll back grpc (#368)"

This reverts commit c01ef95d04.
2016-08-15 11:02:54 -07:00
Wapaul1
362ffa1f3c Changing hard coded ports for objstore and workers to choose unused ports (#365)
* let grpc choose unused worker and object store ports

* Add objstore addresses to scheduler info to bring back test
2016-08-10 19:08:38 -07:00
Robert Nishihara
fbc49410ec properly shutdown worker service thread (#367) 2016-08-10 16:53:24 -07:00
Robert Nishihara
13df8302e6 enable running example apps in cluster mode (#357) 2016-08-08 16:01:13 -07:00
Robert Nishihara
a1e4268d37 Catch errors in importing reusable variables and remote functions (#354)
* catch errors in importing reusable variables and remote functions

* updates
2016-08-07 13:53:33 -07:00
Philipp Moritz
8bf877ac1e Serialize and Deserialize unicode (#349) 2016-08-04 21:06:31 -07:00
Robert Nishihara
ac363bf451 Let worker get worker address and object store address from scheduler (#350) 2016-08-04 17:47:08 -07:00
Johann Schleier-Smith
3ee0fd8f34 Update cluster guide (#347)
* clarify cluster setup instructions

* update multinode documentation, update cluster script, fix minor bug in worker.py

* clarify cluster documentation and fix update_user_code
2016-08-04 09:14:20 -07:00
Robert Nishihara
de200ff912 better error messages when composing remote functions (#339)
Better error messages when composing remote functions
2016-08-03 16:47:13 -07:00
Robert Nishihara
07baf44f26 only display failed task information for new tasks, remove SHELL_MODE (#344) 2016-08-02 17:59:42 -07:00
Robert Nishihara
c27e6c076c Make sure no Python modules mutually import each other. (#334) 2016-08-01 17:55:38 -07:00
Wapaul1
96a70e1316 Changed equality to None in worker.py to use is keyword (#335) 2016-08-01 16:56:54 -07:00
Wapaul1
97b923a750 Changed how ray treats deserialization of custom classes (#333) 2016-08-01 15:38:05 -07:00
Robert Nishihara
98a508d6ca Terminology change Object Reference -> Object ID (#330) 2016-07-31 19:58:03 -07:00
Robert Nishihara
2040372084 unify starting local cluster with attaching to existing cluster (#327) 2016-07-31 19:26:35 -07:00
Robert Nishihara
0e5b858324 change remote function invocation from func() to func.remote() (#328) 2016-07-31 15:25:19 -07:00
Robert Nishihara
352e5e1dd2 use Ray specific logger so logging does not interfere with other python modules that use the logging module (#321) 2016-07-29 12:40:45 -07:00
mehrdadn
4c323d4ecf libraylib -> lib (#311) 2016-07-27 23:23:13 -07:00
Philipp Moritz
dd301e4baa add version string (#304) 2016-07-27 11:29:22 -07:00
Robert Nishihara
975bfe89ad check that we are connected before calling ray commands (#303) 2016-07-27 11:20:37 -07:00
Philipp Moritz
b5215f1e6a make it possible to use directory as user source directory that doesn't contain worker.py (#297) 2016-07-26 18:39:06 -07:00
Robert Nishihara
aa2f618ab7 add directory containing script to python path of workers (#296) 2016-07-26 16:18:39 -07:00
Robert Nishihara
3bae6f136b export remote functions and reusable variables that were defined before connect was called (#292) 2016-07-26 11:40:09 -07:00
Robert Nishihara
8e9f98c5ff Properly import remote functions and reusable variables on workers that register late (#290) 2016-07-25 16:17:17 -07:00
Philipp Moritz
4a0f35b042 new arrow serialization code (serialize python objects recursively) (#284) 2016-07-25 13:41:47 -07:00
mehrdadn
5ff00e0e81 Fix pickling (#289) 2016-07-24 12:53:55 -07:00
Robert Nishihara
03f1830cd0 implement key value store for sharing reusable variables 2016-07-22 18:43:33 -07:00
Robert Nishihara
3bf46ce5ac make remote functions pickleable (#281) 2016-07-19 16:05:45 -07:00