Commit Graph

  • eea19371b7 Suppress warning about working dying when driver exits. (#492) Robert Nishihara 2017-04-26 23:52:13 -07:00
  • 1627f89945 Fix problem in which actors and workers running tasks are not killed by driver exit. (#490) Robert Nishihara 2017-04-26 15:13:39 -07:00
  • b7ace01b5f Convert Plasma client to STL (#486) Philipp Moritz 2017-04-25 01:25:40 -07:00
  • 0ac125e9b2 Clean up when a driver disconnects. (#462) Robert Nishihara 2017-04-24 18:10:21 -07:00
  • 8194b71f32 Convert pending_notifications to STL (#484) Philipp Moritz 2017-04-24 14:41:34 -07:00
  • 892e53d69e Convert plasma client array and object notification queue to STL (#482) Philipp Moritz 2017-04-24 00:43:48 -07:00
  • e36de2dad1 Convert object table to STL (#480) Philipp Moritz 2017-04-23 22:24:05 -07:00
  • 3a2eb1467b Fix failure to propagate error message. (#479) Robert Nishihara 2017-04-23 16:12:25 -07:00
  • a67a107e0e Fix int-type compilation problem on redhat. (#472) Alexey Tumanov 2017-04-19 02:43:33 -07:00
  • cf68cf743c Change UniqueID hash function to look at the lowest instead of highest bytes. (#469) Richard Shin 2017-04-18 15:31:49 -07:00
  • 8ac6c59931 Remove n^2 algorithm in plasma get (#466) Philipp Moritz 2017-04-17 23:37:33 -07:00
  • 0189b09581 Fixes Mac OSX installation error (#464) Guru Medasani 2017-04-16 17:02:15 -05:00
  • c802e51d36 Re-enable recursive remote functions in a limited form. (#453) Robert Nishihara 2017-04-13 01:47:33 -07:00
  • dad57e3b62 Convert actor data structures to C++. (#454) Robert Nishihara 2017-04-12 01:18:16 -07:00
  • 94f32db5e6 A3C Polishing (#385) Richard Liaw 2017-04-11 22:51:52 -07:00
  • c3a2505ffd Loadbalancing Test issue (#452) Richard Liaw 2017-04-11 22:33:58 -07:00
  • fb4525f833 Convert some local scheduler data structures to C++ STL. (#445) Robert Nishihara 2017-04-10 21:02:36 -07:00
  • 6ffc849d23 Use Arrow Tensors for serializing numpy arrays and get rid of extra memcpy. (#436) Philipp Moritz 2017-04-10 01:37:34 -07:00
  • c9d66555e2 Fix bug in queue_task function in local scheduler. (#443) Robert Nishihara 2017-04-09 19:34:43 -07:00
  • f4c1adae17 Unify function signature handling between remote functions and actor … (#441) Robert Nishihara 2017-04-08 21:34:13 -07:00
  • b6c4ae82c0 Increase redis client pubsub buffer size. (#442) Alexey Tumanov 2017-04-08 15:24:07 -07:00
  • 7cd00741b1 Suppress irrelevant Redis connection errors. (#434) Robert Nishihara 2017-04-07 23:19:24 -07:00
  • 0eac3ccdd0 Reduce verbosity of component_failures_test.py. (#440) Robert Nishihara 2017-04-07 23:05:29 -07:00
  • 05fd4c2c37 Changes to local scheduler client protocol. (#435) Robert Nishihara 2017-04-07 23:03:37 -07:00
  • 8061b3b596 Revert "Suppress warning in start_ray.sh about leaving child processes running when parent exits. (#429)" (#437) Robert Nishihara 2017-04-07 17:32:28 -07:00
  • 6f9225490b Plasma manager performance: speed up wait with a wait request object map (#427) Alexey Tumanov 2017-04-07 12:32:12 -07:00
  • 7af6f462fb Add API for querying global control state. (#431) Robert Nishihara 2017-04-06 23:51:12 -07:00
  • 320109a5bd By default, start a number of workers equal to the number of CPUs. (#430) Robert Nishihara 2017-04-06 00:02:58 -07:00
  • fa363a5a3a Notify driver when a worker dies while executing a task. (#419) Robert Nishihara 2017-04-06 00:02:39 -07:00
  • 85b373a4be Suppress warning in start_ray.sh about leaving child processes running when parent exits. (#429) Robert Nishihara 2017-04-05 23:54:22 -07:00
  • 93679df724 Stopped nodes can rejoin immediately (#428) Stephanie Wang 2017-04-05 23:50:38 -07:00
  • 4043769ba2 Make putting large objects work. (#411) Philipp Moritz 2017-04-05 01:04:05 -07:00
  • 1e84747e13 Remove incorrect check. (#421) Robert Nishihara 2017-04-03 14:51:53 -07:00
  • 227c916c25 Convert plasma/plasma_store.cc to use STL (#324) Richard Shin 2017-04-01 13:58:10 +08:00
  • f1b48f2fd4 Avoid publishing in the task table unnecessarily. (#416) Robert Nishihara 2017-03-30 13:41:32 -07:00
  • 036b873bf2 Implement local scheduler task queues using C++ data structures (#392) Stephanie Wang 2017-03-30 00:40:01 -07:00
  • 8245758ccb Add overview of internals to documentation, improve serialization doc… (#390) Robert Nishihara 2017-03-27 21:52:17 -07:00
  • 78e1167a42 Parallelize make in build.sh. (#371) Alexey Tumanov 2017-03-27 20:55:50 -07:00
  • 0925e11c48 Exclude function source from function ID hash in Python interpreter. (#395) Robert Nishihara 2017-03-25 11:31:21 -07:00
  • 054a046b69 Fix installation instructions on Ubuntu and convert md -> rst. (#389) Robert Nishihara 2017-03-24 17:33:26 -07:00
  • a3d58607bf parallelize numbuf memcpy and plasma object hash construction (#366) Alexey Tumanov 2017-03-21 16:17:35 -07:00
  • ba02fc0eb0 Run flake8 in Travis and make code PEP8 compliant. (#387) Robert Nishihara 2017-03-21 12:57:54 -07:00
  • 083e7a28ad Push an error to the driver when the workload hangs on ray.put reconstruction (#382) Stephanie Wang 2017-03-21 00:16:48 -07:00
  • 4618fd45b1 Port Ray to latest Arrow version (#370) Philipp Moritz 2017-03-20 16:31:46 -07:00
  • 29c8471fd4 Add multinode tests by simulating multiple nodes using Docker. (#378) Johann Schleier-Smith 2017-03-18 23:44:54 -07:00
  • 6d9820ef5d Added tensorboard to resnet (#374) Wapaul1 2017-03-17 18:36:23 -07:00
  • 12c9618c0c Plasma and worker node failure. (#373) Stephanie Wang 2017-03-17 17:03:58 -07:00
  • 964d5cac48 Expand API documentation. (#375) Robert Nishihara 2017-03-17 16:48:25 -07:00
  • 6b1e8caf2d Reduce stress_test verbosity. (#377) Robert Nishihara 2017-03-16 20:10:56 -07:00
  • f1d4dda8cb Put all log files in redis and visualize them in UI. (#350) Robert Nishihara 2017-03-16 15:27:00 -07:00
  • 3333e1d6b9 Fix bug in parsing of tasks in monitor. (#372) Robert Nishihara 2017-03-15 20:32:23 -07:00
  • 068429ffd8 Convert local scheduler messages to flatbuffers (#340) Philipp Moritz 2017-03-15 16:27:52 -07:00
  • 4af0aa6258 Atari on pixels (#364) Philipp Moritz 2017-03-14 13:31:29 -07:00
  • 99583f5b08 Clean up rl_pong example. (#365) Robert Nishihara 2017-03-11 21:16:36 -08:00
  • ced13ca5b1 Error Messages - UI display (#360) Richard Liaw 2017-03-11 18:43:06 -08:00
  • b1cb48159a Examples updated with actors. (#358) Wapaul1 2017-03-11 15:30:31 -08:00
  • 3b7788bf88 Disallow calling ray.put on an object ID. (#353) Robert Nishihara 2017-03-11 12:09:28 -08:00
  • b463d9e5c7 Initial A3C Example - PongDeterministic-v3 (#331) Richard Liaw 2017-03-11 00:57:53 -08:00
  • 53dffe0bf2 Use flatbuffers for some messages from Redis. (#341) Robert Nishihara 2017-03-10 18:35:25 -08:00
  • 555dcf35a2 Add policy gradient example. (#344) Philipp Moritz 2017-03-07 23:42:44 -08:00
  • 0de57be085 upgrade flatbuffers to 1.6.0 (#345) Philipp Moritz 2017-03-07 21:33:46 -08:00
  • d001a50644 Add link to the code for the resnet example. (#343) Robert Nishihara 2017-03-07 13:14:00 -08:00
  • c66178bcd7 Resnet Adapted to Ray (#229) Wapaul1 2017-03-07 01:07:32 -08:00
  • da06b4db82 Warn the user when a nondeterministic task is detected. (#339) Stephanie Wang 2017-03-07 00:32:15 -08:00
  • 0b8d279ef2 Convert task_spec to flatbuffers (#255) Philipp Moritz 2017-03-05 02:05:02 -08:00
  • 65a8659f3d Some plasma manager transfer optimizations. (#334) Robert Nishihara 2017-03-04 23:15:17 -08:00
  • 95bf81aeb8 Add actor tutorial. (#335) Robert Nishihara 2017-03-04 23:06:02 -08:00
  • a7ddac6fb1 Properly mock ray submodules when building documentation. (#337) Robert Nishihara 2017-03-04 23:02:56 -08:00
  • 0a233b7144 Update hyperparameter optimization example. (#332) Robert Nishihara 2017-03-04 10:45:15 -08:00
  • 41b8675d04 Availability after local scheduler failure (#329) Stephanie Wang 2017-03-02 19:51:20 -08:00
  • 4f9e74469e Fix segfault induced by getting more than 200k objects (#333) Alexey Tumanov 2017-03-02 01:35:10 -08:00
  • 6a4bde54dc Only install ray python packages. (#330) Robert Nishihara 2017-03-01 23:34:44 -08:00
  • 39b7abefc5 Fix test failures in actor_test.py. (#317) Robert Nishihara 2017-03-01 23:26:39 -08:00
  • 793a102846 Make Ray code C++ compatible (#321) Philipp Moritz 2017-03-01 01:17:24 -08:00
  • ad4b03bf7f Docker Updates (#308) Johann Schleier-Smith 2017-02-28 18:57:51 -08:00
  • b91d9cba45 Adding flatbuffers and migrating flatcc to flatbuffers for plasma (#325) Alexey Tumanov 2017-02-28 18:47:40 -08:00
  • 1a997ed279 Move documentation to ReadTheDocs. (#326) Robert Nishihara 2017-02-27 21:14:31 -08:00
  • 1ae7e7d29e Rename photon -> local scheduler. (#322) Robert Nishihara 2017-02-27 12:24:07 -08:00
  • a30eed452e Change type naming convention. (#315) Philipp Moritz 2017-02-26 00:32:43 -08:00
  • be1618f041 Availability after worker failure (#316) Stephanie Wang 2017-02-25 20:19:36 -08:00
  • 232601f90d Change all table calls to use default retry behavior. (#312) Robert Nishihara 2017-02-24 12:41:32 -08:00
  • aa174e6311 Fix global scheduler test failure. (#314) Robert Nishihara 2017-02-24 11:05:45 -08:00
  • 7f5be96683 Remove object table tests that are failing. (#310) Robert Nishihara 2017-02-23 13:39:59 -08:00
  • 3159a78ad7 terminate photon task dispatch early when workers or resources are unavailable (#311) Alexey Tumanov 2017-02-23 00:05:16 -08:00
  • 54238c4ad0 Propagate errors from importing actors. (#309) Robert Nishihara 2017-02-22 15:15:45 -08:00
  • a6bf16f6a9 Make global scheduler periodically resubmit tasks that can't be sched… (#306) Robert Nishihara 2017-02-21 23:15:46 -08:00
  • e399f57e6b Let actors use GPUs. (#302) Robert Nishihara 2017-02-21 01:13:04 -08:00
  • 3e67d28922 Address numbuf compiler warnings. (#300) Robert Nishihara 2017-02-20 22:42:03 -08:00
  • 334aed9fa9 Fetch the object after requesting reconstruction during ray.get (#301) Stephanie Wang 2017-02-20 21:41:34 -08:00
  • 2220a33b62 In UI, add timing information for tasks and show cluster scheduling. (#297) Robert Nishihara 2017-02-19 15:12:08 -08:00
  • 124baa7472 Fix bug in redis module tests. (#292) Robert Nishihara 2017-02-18 00:55:57 -08:00
  • abd9987e3b Fix unreliable actor test. (#295) Robert Nishihara 2017-02-18 00:51:08 -08:00
  • 67c591c33b Retry connections in photon connect, consolidate code in io.c (#294) Stephanie Wang 2017-02-17 23:41:21 -08:00
  • 9973a6e37c fix bug in numbuf serialization (#296) Philipp Moritz 2017-02-17 23:35:41 -08:00
  • a0dd3a44c0 Dynamically grow worker pool to partially solve hanging workloads (#286) Stephanie Wang 2017-02-17 17:08:52 -08:00
  • 0bbf08a4ac Fix test_illegal_put failure in plasma test. (#289) Robert Nishihara 2017-02-17 11:06:25 -08:00
  • c9bc488ee0 Redirect process output to log files (#267) Johann Schleier-Smith 2017-02-16 20:34:45 -08:00
  • dd7e8d9105 Avoid segfaults in arrow if data is too large (#287) Philipp Moritz 2017-02-16 15:16:20 -08:00
  • 88a5b4e77b Simplify imports and exports and provide driver isolation for remote functions. (#288) Robert Nishihara 2017-02-16 11:30:35 -08:00
  • 883f945db4 Updated tfutils to use new op naming (#284) Wapaul1 2017-02-15 17:47:53 -08:00