hiro/ray
1
0
Fork 0
mirror of https://github.com/vale981/ray synced 2025-03-09 21:06:39 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions
8297 commits 227 branches 67 tags 234 MiB
Commit graph

4406 commits

Author SHA1 Message Date
Edward Oakes
346994745a
[serve] Get handle in starlette endpoint constructor instead of lazily (#15066) 2021-04-01 16:07:28 -05:00
Ian Rodney
22c1aeb240
[Tests] Skip autoscaler tests on Windows (#15033) 2021-04-01 10:16:42 -07:00
SangBin Cho
005cff0092
Revert "Revert "[Core] Implement long polling-based pubsub to reduce … (#14909) 2021-04-01 09:03:15 -07:00
Kai Fricke
d33b0e4bc3
[tune] Reconcile placement groups every N seconds to avoid bottlenecks when running many short trials (#15011) 
Closes a release blocking issue
 2021-04-01 17:04:44 +02:00
Hao Chen
3e1a0439b7
Fix concurrent actor starting too many threads. (#14927) 2021-04-01 19:58:18 +08:00
Ameer Haj Ali
e02bd990d8
Move monitor.py to autoscaler/_private directory (#15050) 
Co-authored-by: Ameer Haj Ali <ameerhajali@Ameers-MacBook-Pro.local>
 2021-04-01 07:41:47 +03:00
SangBin Cho
463e9b2ef9
Try fixing it. (#15046) 2021-03-31 16:31:47 -07:00
Stephanie Wang
a86a7a6a98
[core] Cap total memory used by executing tasks' arguments (#15027) 
* Task dependency map

* Pinned args threshold

* Unit test and fix

* no leaks

* update

* update

* remove assertion
 2021-03-31 15:38:40 -07:00
Edward Oakes
126b9a6c14
[serve] Add basic upscaling test using cluster_utils (#15044) 2021-03-31 17:18:02 -05:00
Alex Wu
70f45af541
Deflake test_failure (#15026) 2021-03-31 14:59:38 -07:00
Edward Oakes
4061b72f2e
[serve] Add serve.get_deployment() API (#14953) 2021-03-31 14:57:39 -05:00
Simon Mo
57256b456a
[serve] Make sure test_imported_backend is ran (#15043) 2021-03-31 14:32:45 -05:00
Yi Cheng
4480132229
[core] Integration runtime_env with ray client (#14881) 
* server side ready

* client size

* py

* fix

* up

* format

* add files

* add pyx

* up

* up

* up

* add keys

* format

* update

* format

* add unittests

* add files

* up

* up

* fix

* up

* fix thread issue

* format

* fix

* update proto

* Fix

* format

* fix

* more

* fix conflict

* fix

* fix order

* format

* add

* up

* compiling fix

* lint

* fix

* format

* fix some

* some fix

* fix comment

* test cases

* add test

* comments

* fix name

* format

* fix

* revert gcs-kv

* fix comments

* fix failure

* fix test

* format

* fix timeout

* fix

* fix

* fix

* format

* format

* fix flaky test

Co-authored-by: Yi Cheng <singye888@gmail.com>
 2021-03-31 11:39:34 -07:00
Clark Zinzow
91cf272c2e
[Core] Exit autoscaler with a non-zero exit code upon handling SIGINT/SIGTERM (#14518) 2021-03-31 10:08:02 -07:00
Ian Rodney
32e50b8c67
[Docker] Run docker stop in parallel (#14901) 
* first pass at parallel docker stop

* real impl

* use env var variable

* lint fix
 2021-03-31 08:41:52 -07:00
Edward Oakes
107effb370
[serve] Add tests for reconnecting to cluster with ray client (#15029) 2021-03-31 10:08:12 -05:00
Edward Oakes
12f5e5ab62
[serve] Small cleanup in HTTP proxy (#15028) 2021-03-31 09:18:11 -05:00
Ian Rodney
73fb5d6022
[Autoscaler][Docker] Make disable_shm_size_detection more usable (#14913) 2021-03-30 18:10:09 -07:00
Siyuan (Ryans) Zhuang
3aa39142db
[Core] Remove code paths that run plasma store as a process (#14924) 
* enable plasma store as thread by default

remove unused code path that runs plasma store as a process
 2021-03-30 16:19:03 -07:00
Clark Zinzow
ccb0cdaa35
Revert "skip on windows (#14988)" (#15017) 
This reverts commit fe39c88a57.
 2021-03-30 11:47:39 -05:00
Edward Oakes
c5e7ed5671
Revert "Add support for Python 3.9 (#12613)" (#15003) 
This reverts commit 208cde8d9b.
 2021-03-30 08:38:54 -05:00
Travis Addair
e5caaa7d1f
Fixed Dask on Ray for dask>=2021.3.1 which dropped Python 3.6 (#14991) 
* Fixed Dask on Ray compatibility with dask==2021.3.1 which drops Python 3.6 support

* Lint
 2021-03-29 23:21:58 -07:00
SangBin Cho
4edcaa8870
[Stats] Basic implementation for the the periodic asio stats printing support. (#14982) 
* Basic implementation for the the periodic asio stats printing support.

* hacky way to count grpc stats.

* lint

* Fix an issue.

* Revert the request/reply.
 2021-03-29 21:51:16 -07:00
Edward Oakes
3591c0ea2d
Revert "[minor] improve warning message for Ray. #14949" (#15004) 
This reverts commit c84073f3f4.
 2021-03-29 21:15:22 -07:00
Simon Mo
6b49714c04
[Serve] Add tests for more FastAPI features (#14961) 2021-03-29 17:38:51 -07:00
SangBin Cho
eaf159795b
[Test] Fix memory scheduling flaky test (#14980) 2021-03-29 15:44:26 -07:00
Richard Liaw
c84073f3f4
[minor] improve warning message for Ray. #14949 
Signed-off-by: Richard Liaw <rliaw@berkeley.edu>
 2021-03-29 15:17:32 -07:00
Akash Patel
208cde8d9b
Add support for Python 3.9 (#12613) 2021-03-29 11:57:06 -07:00
Edward Oakes
fe39c88a57
skip on windows (#14988) 2021-03-29 10:06:25 -07:00
Edward Oakes
e79d4cf6f5
[serve] Support setting deployment options via kwargs (#14935) 2021-03-29 11:14:27 -05:00
Amog Kamsetty
95ff342558
[Tune] Wandb API Key File Compatibility with Ray Client (#14942) 2021-03-29 15:39:54 +02:00
dependabot[bot]
68c82b6503
[tune](deps): Bump wandb from 0.10.19 to 0.10.23 in /python/requirements (#14964) 
Bumps [wandb](https://github.com/wandb/client) from 0.10.19 to 0.10.23.

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2021-03-29 15:37:56 +02:00
Siyuan (Ryans) Zhuang
87c79553e9
[Core] Remove code paths that contains plasma store executable (#14950) 
* remove plasma store executable & never used tests

* set default behavior

* fix tests
 2021-03-28 21:22:14 -07:00
Micah Yong
b3089b31f2
[RFC] Ray memory improvements: format and summary (#14520) 
* Better formatting when terminal size doesn't support tabular

* Summary now displays size of reference types

* Add unit conversion support (e.g. b, kb, mb, gb)

* Format and test

* Add ability to specify the number of sorted entries

* Linting

* Clean up group summary, move import defaultdict, comment num entries counter, n

* Format and lint
 2021-03-28 21:03:06 -07:00
Dmitri Gekhtman
dcf41d868c
[autoscaler][Kubernetes] Fix non_terminated_nodes consistency (#14976) 
* Verify pod termination

* deletion-timestamp

* get rid of extra constant
 2021-03-28 14:52:12 -07:00
Frank Luan
cdbaf930ab
[metrics] Fix deserialization warnings for metrics.Counter (#14969) 2021-03-28 09:44:30 -05:00
Edward Oakes
fd4ed3acfe
[serve] Skip failing test_deploy tests on windows (#14957) 2021-03-26 13:51:54 -05:00
SangBin Cho
839cd1e0a2
[Core] Remove unnecessary redis connection (#14511) 
* remove unnecessary stuff.

* test in progress.

* Fix tests.

* lint

* fix.

* Remove tests that were not working properly before.
 2021-03-26 10:29:12 -07:00
Eric Liang
2157021fd3
Refactor object restoration path (#14821) 2021-03-25 22:46:50 -07:00
tchordia
4e66efc532
Update ARCHITECTURE.md (#14889) 
update link
 2021-03-25 12:30:35 -07:00
Edward Oakes
63594c5370
[serve] Rolling updates for redeployments (#14803) 2021-03-25 12:23:08 -05:00
Simon Mo
1fcca07856
[Serve] FastAPI Simple Class Based View (#14858) 2021-03-25 12:21:36 -05:00
Kai Fricke
b366500938
[tune] fix long running release test WIP (#14866) 
- Use placement groups
- Introduce time between checks for failure testing
- Use gloo instead of nccl
 2021-03-25 11:03:22 +01:00
Kai Fricke
84b3c3376b
[tune] document scalability best practices (k8s, scalability thresholds) (#14566) 
Adds a new page and table to document current scalability thresholds in Ray Tune to the documentation.

Co-authored-by: Richard Liaw <rliaw@berkeley.edu>
 2021-03-25 09:54:14 +01:00
SangBin Cho
0004d87194
[Test] Refactor object spilling test (#14861) 
* refactoring done.

* refactoring done.
 2021-03-25 00:46:46 -07:00
architkulkarni
03afaed6e1
[Serve] [Doc] Create top-level page for Calling Endpoints from HTTP and from Python (#14904) 2021-03-24 20:29:24 -05:00
Dmitri Gekhtman
25ebefafc8
[autoscaler][aws][test] Validate current state of subnet-specification (#14859) 
* This PR adds a test that validates that adding head_nodes and worker_nodes fields with subnet data to a multi-node-type config leads to a correct configuration of a security group.
 2021-03-25 01:40:16 +02:00
Yi Cheng
f427801c10
Revert "[core] Fix worker type in python (#14823)" (#14910) 
This reverts commit 9ccf291f4d.
 2021-03-24 13:27:56 -07:00
Simon Mo
d57808d007
[Serve] Add support for handle.method_name.remote (#14831) 2021-03-24 12:10:14 -07:00
Edward Oakes
59e231818d
[serve] Add Deployment.delete() and un-skip relevant tests (#14898) 2021-03-24 13:40:30 -05:00