Commit Graph
4036 Commits
Author SHA1 Message Date
Eric LiangandGitHub 46af992efd [rllib] [experimental] custom RL training pipelines (PG_pl, A2C_pl) (#7213) 2020-02-19 16:07:37 -08:00
Simon MoandGitHub 7bef7031c2 Revert "Revert "Revert "Removing Pyarrow dependency (#7146)" (#7209) (#7214)" (#7232) 2020-02-19 13:35:29 -08:00
Sven MikaandGitHub d537e9f0d8 [RLlib] Exploration API: merge deterministic flag with exploration classes (SoftQ and StochasticSampling). (#7155) 2020-02-19 12:18:45 -08:00
Eric LiangandGitHub 399424c418 [rllib] Fix broken check in eval mode for IMPALA #7217 2020-02-19 11:54:30 -08:00
Simon MoandGitHub e8941b1b79 Revert "Revert "Removing Pyarrow dependency (#7146)" (#7209) (#7214) 2020-02-19 10:08:52 -08:00
f76ce836b2 Distributed ref counting for serialized ObjectIDs (#6945)
* Skeleton plus a unit test for simple borrower case

* First unit test passes - forward an ID and task returns with 1 submitted task pending on the inner ID

* Invariant for contained_in

* Unit test passes for testing task return without creating a borrower

* Wrap ref count functionality in test case

* Fix bad delete

* Unit test and fix for borrowers creating more borrowers

* Unit test and fix for simple borrowing, but owner sends call after borrower's ref count goes to 0

* Refactor:
- keep a sentinel ref count for task argument IDs
- keep contained_in_borrowed in addition to contained_in_owned

* Unit test for nested IDs passes

* Refactor so that an object ID can only be contained in 1 borrowed ID at a time

* Add check

* Fix

* Unit test (passes) to test nesting object IDs but no borrowers created

* Unit test for nested objects from different owners passes, refactor to unset contained_in when popping refs

* Unit tests for borrowers receiving an ObjectID from multiple sources,
skip adding ownership info if we already have it to handle duplicate
refs

* Unit test for returning object ID passes

* More unit tests for returning object IDs pass

* Add serialized ID tests

* fix serialization issue

* remove swap

* It builds!

* debugging and some fixes:
- register handler for WaitForRefRemoved
- don't create a python reference for arg IDs
- pass in client factory into ReferenceCounter
- fix bad decrement in PopBorrowerRefs

* Fix accounting for serialized IDs:
- don't decrement for IDs on dependency resolution, wait until task finished
- add object IDs that were inlined when building the arguments to the task spec, pin these on the task executor until task finishes

* mu_ -> mutex_

* lint

* fix build

* clear outer_object_id

* add direct call type check

* Fix test for direct call IDs and return IDs for actor calls

* Fix CoreWorkerClient.Addr()

* Remove unneeded lock

* Remove unnecessary ObjectID refs

* Fix worker holding serialized refs test

* Fix hex IDs

* fix

* fix tests

* fix tests

* refactor and cleanups

* lint

* Put inlined Ids in task args and some cleanup

* Add back gc.collect() line for test case

* Refactor and fixes:
- store inlined IDs in RayObject
- allow storing objects with inlined IDs in memory store
- pin objects that were promoted to plasma

* oops

* make sure worker ID is set in address, pass in rpc::Address to CoreWorkerClient

* todos

* cleanups and test builds

* Fix tests

* Add feature flag

* cleanups

* address comments and some cleanups

* cleanup

* fix recursive test

* Comments for tests

* Turn off ref counting by default

* Skip tests

* Fix some bugs for test_array.py, java build

* Don't include nested objects in the ref count when the feature flag is off

* C++ feature flag does not work...

* Remove

* Turn on python tests and add a warning when plasma objects are evicted before being pinned

* Fix build and remove irrelevant test

* Fix for java

* Revert "Fix build and remove irrelevant test"

This reverts commit 056cca9b263ed05b0f9ab2250907338edcbca2d5.

* Fix ray.internal.free

* Fixes and skip some flaky tests

* fix java build

* fix windows build

* Add IDs contained in owned objects

* Update src/ray/protobuf/core_worker.proto

Co-Authored-By: Edward Oakes <ed.nmi.oakes@gmail.com>

* Update src/ray/core_worker/reference_count.cc

Co-Authored-By: Edward Oakes <ed.nmi.oakes@gmail.com>

* Update src/ray/protobuf/core_worker.proto

Co-Authored-By: Edward Oakes <ed.nmi.oakes@gmail.com>

* Update src/ray/protobuf/core_worker.proto

Co-Authored-By: Edward Oakes <ed.nmi.oakes@gmail.com>

* Update src/ray/core_worker/reference_count.h

Co-Authored-By: Edward Oakes <ed.nmi.oakes@gmail.com>

* Update src/ray/core_worker/reference_count.h

Co-Authored-By: Edward Oakes <ed.nmi.oakes@gmail.com>

* Update src/ray/core_worker/reference_count.cc

Co-Authored-By: Edward Oakes <ed.nmi.oakes@gmail.com>

* Apply suggestions from code review

Co-Authored-By: Edward Oakes <ed.nmi.oakes@gmail.com>

* update

* Try to fix ::test_direct_call_serialized_id_eviction

Co-authored-by: Edward Oakes <ed.nmi.oakes@gmail.com>
2020-02-18 18:21:34 -08:00
mehrdadnandGitHub 4a12243336 Use Process instead of pid_t (round 2) (#6882)
* Revert "Revert "Use Boost.Process instead of pid_t (#6510)" (#6909)"

This reverts commit bde575b8dd.

* Process wrapper, using Boost.Process on Windows

- Reverts bde575b8dd.
- Re-applies fb8e3615d5 after some refactoring.

* Remove Boost.Process dependency

* Don't open /proc file on Linux

* Change FATAL to ERROR and modify error message when process doesn't exist
2020-02-18 17:44:46 -08:00
Eric LiangandGitHub 0aa9373d62 Revert "Removing Pyarrow dependency (#7146)" (#7209)
This reverts commit 2116fd3bca.
2020-02-18 14:12:06 -08:00
Eric LiangandGitHub 5df801605e Add ray.util package and move libraries from experimental (#7100) 2020-02-18 13:43:19 -08:00
Eric LiangandGitHub fae99ecb8e [core] Make sure to unsubscribe get dependencies for direct task calls. (#7201)
* fix

* remove assert
2020-02-17 18:35:25 -08:00
ijrsvtandGitHub 2116fd3bca Removing Pyarrow dependency (#7146) 2020-02-17 18:00:13 -08:00
mehrdadnandGitHub 3bd82d0bcd Fix various issues/warnings that come up on Jenkins (#7147)
* Avoid warning about swap being unlimited

Currently we get the following message on Jenkins:
"Your kernel does not support swap limit capabilities or the cgroup is not mounted. Memory limited without swap."

Since we're not limiting swap anyway, we might as well avoid trying to.
https://docs.docker.com/config/containers/resource_constraints/#--memory-swap-details

* Fix escaping in re.search()

* Fix escaping in _noisy_layer()

* Raise a more descriptive error when dashboard data isn't found

* Don't error on dashboard files not being found when webui isn't required

* Change dashboard error to a warning instead
2020-02-17 16:08:55 -08:00
Alex WuandGitHub 734629b4ea Ssh command format (#7176) 2020-02-17 14:15:42 -08:00
Alind KhareandGitHub c6d768be14 [Serve] Added support for no http route services (#7010) 2020-02-17 11:31:30 -08:00
Eric LiangandGitHub 42aea966ff [rllib] Convert torch state arrays to tensors during compute actions (#7162)
* convert to tensor

* normalize fix
2020-02-17 10:26:58 -08:00
fyrestoneandGitHub a6b8bd47b0 [xlang] Cross language serialize ActorHandle (#7134) 2020-02-17 20:44:56 +08:00
Edward OakesandGitHub b079787c59 Fix flaky test_get_with_timeout (#7175) 2020-02-16 21:10:16 -08:00
94e2fcea2e [sgd] fp16 (apex) and scheduler support + move examples page (#7061)
* Init fp16

* fp16 and schedulers

* scheduler linking and fp16

* to fp16

* loss scaling and documentation

* more documentation

* add tests, refactor config

* moredocs

* more docs

* fix logo, add test mode, add fp16 flag

* fix tests

* fix scheduler

* fix apex

* improve safety

* fix tests

* fix tests

* remove pin memory default

* rm

* fix

* Update doc/examples/doc_code/raysgd_torch_signatures.py

* fix

* migrate changes from other PR

* ok thanks

* pass

* signatures

* lint'

* Update python/ray/experimental/sgd/pytorch/utils.py

* Apply suggestions from code review

Co-Authored-By: Edward Oakes <ed.nmi.oakes@gmail.com>

* should address most comments

* comments

* fix this ci

* fix tests'

* testmode

Co-authored-by: Edward Oakes <ed.nmi.oakes@gmail.com>
2020-02-16 19:04:08 -08:00
Sven MikaandGitHub f0e62d733f Bazel exclude rllib-option fix. (#7185) 2020-02-16 11:26:03 -08:00
Eric LiangandGitHub b7016504e8 [rllib] Only run one set of tests unless rllib or tune dirs are changed. (#7179)
* full filter

* lint
2020-02-16 08:52:49 -08:00
Siyuan (Ryans) ZhuangandGitHub 6745459f96 Apply cpython patch bpo-39492 for the reference counting issue in pickle5 (#7177)
* apply cpython patch bpo-39492 for the reference count issue
2020-02-15 21:16:13 -08:00
Eric LiangandGitHub b6233dff3c [rllib] Fix bad sample count assert 2020-02-15 17:22:23 -08:00
Sven MikaandGitHub 2e60f0d4d8 [RLlib] Move all jenkins RLlib-tests into bazel (rllib/BUILD). (#7178)
* commit

* comment
2020-02-15 14:50:44 -08:00
Edward OakesandGitHub dc5a27dac0 Move ray.experimental.multiprocessing to ray.util.multiprocessing (#7149) 2020-02-14 16:17:05 -08:00
Richard LiawandGitHub 52d9189d5d [autoscaler] port-forward for attach + redis_port (#7145)
* port-forward

* fixport

* force redis port in init mode

* test

* Update python/ray/tests/test_ray_init.py
2020-02-14 15:17:00 -08:00
Simon MoandGitHub 30de1286bd Use pip install setup.py (#7158) 2020-02-14 13:53:36 -08:00
Adrian O'GradyandGitHub fe6ce714a0 [rllib] - TaskPool.completed_prefetch() no longer returns stale object ids after an error (#7139) 2020-02-13 22:30:44 -08:00
Qing WangandGitHub f3703bafa3 [Java] Support concurrent actor calls API. (#7022)
* WIP

Temp change

Attach native thread to jvm

* Fix run mode

* Address comments.
2020-02-14 13:02:39 +08:00
Alex WuandGitHub 0d3687a10d No warning for docker memory > system memory (#7151) 2020-02-13 15:21:44 -08:00
Edward OakesandGitHub b81b93a9c0 Convert stress tests to projects (#6495) 2020-02-13 09:19:24 -08:00
94a286ef1d [Java] Add session_dir as temp_dir for logs, socket files like Python (#7044)
* Support

* Add gcs_server support

* Fix ut

* Fix

* Remove unused py code

* Fix linting

* Fix cross language ci

* Fix CI

* Add docstring

* Fix

* Fix linting

* Add a singleton for config

* Refine

* fix

* Fix

* linting

* Remove FileUnit

* Fix

* Fix

* Fix

* Update java/runtime/src/main/java/org/ray/runtime/config/RayConfig.java

Co-Authored-By: Hao Chen <chenh1024@gmail.com>

* Fix streaming singleprocess CI

* Fix checkstyle

Co-authored-by: Hao Chen <chenh1024@gmail.com>
2020-02-13 17:49:52 +08:00
Sven MikaandGitHub 5518a738b3 [RLlib] Fix erroneous use of LinearSchedule (in DDPG's exploration annealing). (#7125)
* Fix erroneous use of LinearSchedule (in DDPG's exploration annealing).
Erase schedules_obsoleted.py.

* Trigger re-test.

* Re-test.
2020-02-12 23:46:49 -08:00
wanxingandGitHub 9fc3e2e50f [Streaming]Add RefreshChannelInfo to support flow-control (#7071)
* add RefreshChannelInfo

* fix name

* add override

* fix

* fix return value
2020-02-13 09:30:56 +08:00
Edward OakesandGitHub e904711e74 Add python tests for serialized object ID reference counting (#7038) 2020-02-12 16:52:07 -08:00
Edward OakesandGitHub d91d3ea936 Split half of test_actor into test_actor_advanced (#7143) 2020-02-12 15:17:25 -08:00
Simon MoandGitHub 0e94e1dc2a [Asyncio] Increase recursion limit manually (#7142) 2020-02-12 14:15:36 -08:00
Sven MikaandGitHub f41a9b9813 [RLlib] Fix KL method of MultiCategorial tf distribution (issue #7009). (#7119)
* Fix KL method of MultiCategorial tf distribution.

* Fix KL method of MultiCategorial tf distribution.

* Merge AsyncReplayOptimizer fixes into this branch.
2020-02-12 12:46:15 -08:00
Edward OakesandGitHub 275fd343fb Change CI to properly list python3.6 (#7126) 2020-02-12 11:15:46 -08:00
Mitchell SternandGitHub 5dda0b66bf [Dashboard] Refactor dialogs to use parent component state instead of routes (#7129) 2020-02-12 10:59:47 -08:00
aannadiandGitHub d941ac6c89 Updating package-lock.json with latest npm (#7128) 2020-02-12 09:54:20 -08:00
fc9352c588 [docs] Make walkthrough and starting Ray materials clear (#7099)
* make starting ray a separate page

* concept

* Apply suggestions from code review

Co-Authored-By: Edward Oakes <ed.nmi.oakes@gmail.com>

* more fics

* Apply suggestions from code review

Co-Authored-By: Edward Oakes <ed.nmi.oakes@gmail.com>

Co-authored-by: Edward Oakes <ed.nmi.oakes@gmail.com>
2020-02-11 23:17:30 -08:00
Eric LiangandGitHub 305eaaabe9 Fix hang if actor object id is returned from a task that exits (#6885) 2020-02-11 20:28:13 -08:00
mehrdadnandGitHub e09f63ad65 Fix build errors and add more targets to Windows builds (#6811)
* Fix common.fbs rename (due to apache/arrow/commit/bef9a1c251397311a6415d3dc362ef419d154caa)

* Add missing COPTS

* Use socketpair(AF_INET) if boost::asio::local is unavailable (e.g. on Windows)

* Fix compile bug in service_based_gcs_client_test.cc (fix build breakage in #6686)

* Work around googletest/gmock inability to specify override to avoid -Werror,-Winconsistent-missing-override

* Fix missing override on IsPlasmaBuffer()

* Fix missing libraries for streaming

* Factor out install-toolchains.sh

* Put some Bazel flags into .bazelrc

* Fix jni_md.h missing inclusion

* Add ~/bin to PATH for Bazel

* Change echo $$(date) > $@ to date > $@

* Fix lots of unquoted paths

* Add system() call checks for Windows

Co-authored-by: GitHub Web Flow <noreply@github.com>
2020-02-11 16:49:33 -08:00
Simon MoandGitHub 039d2cde88 Change log level for OMP warning (#7114) 2020-02-11 14:15:38 -08:00
Sven MikaandGitHub 2a0e4d94aa [RLlib] Fix AsyncReplayOptimizer bug where it swallows all good worker tasks … (#7111) 2020-02-11 12:51:44 -08:00
mehrdadnandGitHub fea54ab97f Use --force_pic to prevent Bazel from compiling the same files twice, which slows down the build considerably. (#7116) 2020-02-11 12:11:53 -08:00
aannadiandGitHub d7ff55852a [tune][Dashboard] Added Tune Dashboard (#6911) 2020-02-11 11:56:49 -08:00
Simon MoandGitHub 0d5225c743 [Build] Use latest known tagged arrow image for linux build (#7115) 2020-02-11 11:17:48 -08:00
Simon MoandGitHub 0ddc389830 Fix documentation building with psutil issue (#7077) 2020-02-11 10:00:29 -08:00
Lingxuan ZuoandGitHub 3d9bd64591 [Streaming] DataWriter use event driven model. (#7043)
* streaming writer use event driven model.

* minor changes according reviewer comments

* Fix according to reviewer's comments

* fix bazel lint

* code polished

* Add more comments

* rename Stop & Start of EventQueue to Freeze and Unfreeze.
2020-02-11 22:24:45 +08:00