Commit graph

4594 commits

Author SHA1 Message Date
Antoni Baum
0935ec30d0
[tune] Add information about environment variables to tune.run docstring (#16980) 2021-07-11 17:20:17 -07:00
Julius Frost
a88b217d3f
[rllib] Enhancements to Input API for customizing offline datasets (#16957)
Co-authored-by: Richard Liaw <rliaw@berkeley.edu>
2021-07-10 15:05:25 -07:00
Scott Graham
3334357c58
[autoscaler] [azure] Fix Azure Autoscaling Failures (#16640)
Co-authored-by: Scott Graham <scgraham@microsoft.com>
2021-07-10 11:55:00 -07:00
Nikita Vemuri
6d36d7ed7e
[Serve] Call FastAPIWrapper class constructor before startup hooks (#16941)
* run constructor before startup hooks

* address comments

Co-authored-by: Nikita Vemuri <nikitavemuri@nikitas-mbp.attlocal.net>
2021-07-09 09:39:32 -07:00
Dmitri Gekhtman
27a9ae5e13
[autoscaler][gcp] Retry GCP BrokenPipeError (#16952) 2021-07-08 13:54:29 -07:00
Maxim Egorushkin
9cb5c9a422
Never convert trial_id to float when loading progress.csv. (#16959)
* Never convert trial_id to float when loading progress.csv.

* Formatting updated.

Co-authored-by: Maxim Egorushkin <maxim.egorushkin@gmail.com>
2021-07-08 11:06:11 -07:00
SongGuyang
560fd15568
[C++ worker] support build and add C++ worker to python wheel (#16496) 2021-07-08 14:42:26 +08:00
Clark Zinzow
cc215353e2
[Datasets] Adds Dataset.iter_batches(). (#16853) 2021-07-07 22:01:20 -07:00
Frank Luan
7c0320175c
Actor fix (#16955) 2021-07-07 20:51:36 -07:00
Clark Zinzow
9358dd4bc2
[Datasets] Port JSON and CSV readers to datasource API. (#16938)
* Port JSON and CSV readers to datasource API.

* Formatting.

* Moved datasources to datasource dir, created shared FileBasedDatasource.

* Confirm that accessing dataset schema raises an error.

* Formatting.

* Return None for unknown metadata instead of raising an error.

* Feedback.
2021-07-07 20:32:04 -07:00
Kai Yang
e925051ce4
[Core] Get node to connect for driver in global state accessor (#16810) 2021-07-08 11:21:12 +08:00
Amog Kamsetty
3c482cd6c8
Skip more test_deploy tests on OSX (#16943)
* skip more

* skip more
2021-07-07 16:53:21 -07:00
Simon Mo
f4671d55d8
Bump log monitor's sleep duration to 0.1s (#16939)
We observed in long running serving scenarios the log monitor
consistently uses 10% of cpus when there is no new lines. Hopefully
this new sleep duration should shrink that usage
2021-07-07 15:41:34 -07:00
Chen Shen
0421fa188e
[core] use fallocate for fallback allocation to avoid SIGBUS (#16824) 2021-07-07 14:50:11 -07:00
Dmitri Gekhtman
2f42b0c4b9
[kubernetes] K8s keep gpu zero override (#16887) 2021-07-07 13:45:34 -07:00
Chen Shen
dbd3260141
[core] Deprecate QuotaAwareEvictionPolicy (#16911) 2021-07-07 13:44:41 -07:00
Eric Liang
3b9f6ccc5e
Remove autoinit from ray.data (#16925) 2021-07-07 13:44:10 -07:00
Amog Kamsetty
b79ef3ba0f
[Serve] Skip more test_deploy tests on OSX (#16937) 2021-07-07 10:44:01 -07:00
Antoni Baum
8f41a34079
[tune] Placement group manager fixes (#16844)
Co-authored-by: Kai Fricke <kai@anyscale.com>
2021-07-07 10:42:19 -07:00
Antoni Baum
b737b2a877
[joblib] Improved object store management for Pool (#16879)
* Improved object store management for Pool

* Update docs, hints

* Add test

* Nit

* Nit
2021-07-07 10:39:18 -07:00
Dmitri Gekhtman
c6497c6520
[client][test] Client multiprocessing tests + client api minor fix (#16904) 2021-07-07 09:47:27 -07:00
Eric Liang
03f99100ea
Enable ray auto init by default (#16861) 2021-07-06 21:56:32 -07:00
Eric Liang
ca083e16d4
[dataset] Fix conversion to pyarrow tables in several transforms (#16916) 2021-07-06 20:40:57 -07:00
Amog Kamsetty
7318a212fb
[Serve] Skip test_redeploy_multiple_replicas on OSX (#16915) 2021-07-06 18:58:36 -07:00
Eric Liang
7e52fde8a3
Fix num returns error message (#16865) 2021-07-06 14:57:26 -07:00
Stefan Schneider
d4babd69c1
[windows] correct symlinks for files (node.py) (#16817) 2021-07-06 10:01:13 -07:00
Dmitri Gekhtman
a27a8172cc
[autoscaler] Handle node type key change/deletion (#16691) 2021-07-06 09:06:58 -07:00
Kai Fricke
4178655ba7
[tune] Pass custom sync_to_cloud templates to durable trainables (#16739) 2021-07-06 09:50:59 +01:00
Eleven Liu
e250abf689
[tune] Sort top results by metric (#16576) 2021-07-06 08:59:31 +01:00
Eric Liang
4af36faea1
[docs] Cleanup workflow api.py pydoc and spell out ObjectRef for clarity (#16857)
* cleanup types

* docs

* clarify
2021-07-06 00:59:06 -07:00
Kai Yang
7c21be5450
[Object spilling] Clean up spilled objects on disk when Raylet starts (#16669) 2021-07-05 12:01:25 +08:00
Vince Jankovics
63ce4b4e97
[tune] Fix step for MLflow log_trial_result (#16840)
* Fix step for MLflow log_trial_result

* fix test

* lint

Co-authored-by: Amog Kamsetty <amogkamsetty@yahoo.com>
2021-07-03 10:06:45 -07:00
Yi Cheng
4bb3883a73
[dataset] deduct filesystem automatically (#16762) 2021-07-03 00:50:59 -07:00
Siyuan (Ryans) Zhuang
122bf309fa
[Workflow] Workflow management - Part I (#16838)
* refactoring

* share fate with the driver

* move TODOs to correct locations

* disable objectref test

* test raise exception when use object ref as inputs
2021-07-02 22:12:45 -07:00
SangBin Cho
61451af06b
[OBOD] Bug fix from test_scheduling.py (#16791) 2021-07-02 19:26:31 -07:00
architkulkarni
f02e41a822
[Core] [runtime env] Add RuntimeEnvHash and JobID to SchedulingKey (#16766)
* add python integration test

* improve readability

* remove unneccessary ray start --head

* add shutdown_only

* move RuntimeEnvHash from worker_pool to task_spec

* lint

* Add runtimeEnvHash and JobID to SchedulingKey

* remove JobID from key and hopefully fix compile

* add test for same env

* lint
2021-07-02 18:15:28 -07:00
Eric Liang
f2f6176057
Deflake test_failure_3 (#16854) 2021-07-02 18:00:42 -07:00
loicsacre
b7be848f07
[tune] Allow the logging of wandb.Image in tune.integration.wandb (#16839) 2021-07-02 17:14:23 -07:00
Charles
2526299eba
[Autoscaler] Use NetworkInterfaces parameter to launch instances. (#16816) 2021-07-02 14:01:02 -07:00
dependabot[bot]
ba13c2f7dd
[RLlib](deps): Bump gym[atari] in /python/requirements/rllib (#15992)
Bumps [gym[atari]](https://github.com/openai/gym) from 0.18.0 to 0.18.3.
- [Release notes](https://github.com/openai/gym/releases)
- [Commits](https://github.com/openai/gym/compare/0.18.0...0.18.3)

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2021-07-02 09:58:18 -07:00
dependabot[bot]
ee9a1b022e
[tune](deps): Bump kubernetes in /python/requirements/tune (#16002)
Bumps [kubernetes](https://github.com/kubernetes-client/python) from 12.0.1 to 17.17.0.
- [Release notes](https://github.com/kubernetes-client/python/releases)
- [Changelog](https://github.com/kubernetes-client/python/blob/master/CHANGELOG.md)
- [Commits](https://github.com/kubernetes-client/python/compare/v12.0.1...v17.17.0)

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2021-07-02 09:48:02 -07:00
dependabot[bot]
81866a754a
[tune](deps): Bump optuna in /python/requirements/tune (#16376)
Bumps [optuna](https://github.com/optuna/optuna) from 2.6.0 to 2.8.0.
- [Release notes](https://github.com/optuna/optuna/releases)
- [Commits](https://github.com/optuna/optuna/compare/v2.6.0...v2.8.0)

---
updated-dependencies:
- dependency-name: optuna
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2021-07-02 09:46:52 -07:00
dependabot[bot]
0959fbca96
[tune](deps): Bump transformers in /python/requirements/tune (#16672)
Bumps [transformers](https://github.com/huggingface/transformers) from 4.3.2 to 4.8.1.
- [Release notes](https://github.com/huggingface/transformers/releases)
- [Commits](https://github.com/huggingface/transformers/compare/v4.3.2...v4.8.1)

---
updated-dependencies:
- dependency-name: transformers
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2021-07-02 09:45:23 -07:00
Kai Fricke
6c29f2bed6
[tune] Use fork for wandb, disable ipython in trainables (#16675) 2021-07-02 15:52:24 +01:00
Antoni Baum
cabd159423
FIx PlacementGroupFactory hashing (#16842) 2021-07-02 15:44:39 +01:00
Eric Liang
e77a964640
[dataset] Add datasource API (#16826) 2021-07-01 23:44:30 -07:00
Ian Rodney
f9daf7fa2c
[Connect] Separate Specific Server logs into different files (#16534)
* better debug logging

* log to separate files

* use getter/setter methods

* actually use type for ray.node.Node

* update docstring

* use public method

* add a test

* fix glob glob

* fix test again
2021-07-01 21:27:03 -07:00
Kai Fricke
521a05834d
[tune] Fix resume=REMOTE documentation and log more details (#16709) 2021-07-01 18:40:15 -07:00
Eric Liang
7eec197d31
[data] Support block metadata and lazy loading of blocks from the filesystem (#16689) 2021-07-01 17:25:20 -07:00
Siyuan (Ryans) Zhuang
b570606438
[Workflow] Ensure asynchronous execution of "workflow.run" & fix TODOs (#16727)
* initial commit

* code refactoring; non-blocking "workflow.run"

* Rename "workflow_root_dir" to "storage_url"

* readable workflow_id and step_id

* verbose logging

* remove unused code

* slugify

* storage -> storage_or_url

* fix comments

* storage_or_url -> storage

* get workflow outputs without extra object copy & transfer

* update tests

* tests workflow outputs resolving
2021-07-01 15:53:22 -07:00