forked from ray-project/ray
-
Notifications
You must be signed in to change notification settings - Fork 1
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[pull] master from ray-project:master #2320
Merged
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: Matthew Deng <[email protected]>
This PR allows to cancel actor tasks. The details are written in this doc https://docs.google.com/document/d/12avlF2OoFLs8lfC18i9ATXWrapxH1vFMy2UZU7L0xHI/edit (the doc is still in progress).
…ns (#38554) Some of our (internal) class trainables return a subpath in `save_checkpoint`. This path has been deprecated with the new storage refactor. This PR updates the implementations to reconstruct the checkpoint path in `load_checkpoint` and return None in `save_checkpoint` so adhere to the new API. Signed-off-by: Kai Fricke <[email protected]>
…nse` (#37817) Adds new DeploymentHandle that removes the need for "double await" and also unifies the sync & async handle types. See docstrings for detailed API information. The new API can be used either by setting `RAY_SERVE_ENABLE_NEW_HANDLE_API=1` at the cluster level or by using `handle.options(use_new_handle_api=True)`. The migration plan is as follows: - In Ray 2.7, we will be fully backwards-compatible and users will need to opt-in to use the new API (via environment variable or .options() flag). - The documentation and other docstrings will be updated to push users to use the new handle type and warn that the existing one will be deprecated. - Starting in Ray 2.8, we will print deprecation warnings when users use the existing handle API. - In Ray 3.0, the default will be changed to use the new handle.
After Ray 2.5, ActorPoolStrategy() requires using kwargs, otherwise a ValueError will be raised. This updates the example code with keyword arguments. Signed-off-by: Kaining Zhong <[email protected]> Co-authored-by: Amog Kamsetty <[email protected]>
… API (#38617) Signed-off-by: Justin Yu <[email protected]>
Signed-off-by: Lonnie Liu <[email protected]>
…sample in python 311 (#38629) Fix breaking change in Python 3.11 Signed-off-by: rickyyx <[email protected]>
…pyarrow's `read_csv` method (#38496) This PR fixes the issue #35133 Basically, when submitting a ray job to a ray cluster with local file: 'local://LOCAL_FILE_PATH', `reader_args` is passed to pyarrow.csv's `read_csv` method. `reader_args` contain `local_uri` keyword, so that the `{'local_uri':True}` is passed onto pyarrow.csv's `read_csv` where the `read_csv` does not accept `local_uri` keyword. Thus, the PR simply removes the `local_uri` in `reader_args` dict, if one exists. Signed-off-by: Jinsung Ha <[email protected]>
Signed-off-by: Hao Chen <[email protected]>
Migrate core, serve and serverless tests to civ2 Move test runtime builder to wanda to take advantage of cache, one for each team (corebuild, servebuild, serverlessbuild). This is better than do it after ray installation, because otherwise that layer is not cachable. Build ray, and only ray, right before running the tests. ci/ray_ci/tests.env.Dockerfile: basically the minimal of build.Docker part. There are a lot of files, they are all just configurations (or moving things around) Signed-off-by: can <[email protected]>
#38546 tried to fix the issue of missing schema in empty blocks. But that approach doesn't work for ragged tensor arrays. Reverting that change. Per discussion, we'll stop scheduling empty blocks for the map operator instead. Signed-off-by: Hao Chen <[email protected]>
Signed-off-by: Matthew Deng <[email protected]>
Signed-off-by: Kourosh Hakhamaneshi <[email protected]>
Signed-off-by: xwjiang2010 <[email protected]>
…nce (#38678) Signed-off-by: Balaji Veeramani <[email protected]>
add an underscore prefix so that forge building appears on top on the pipeline UI. Signed-off-by: Lonnie Liu <[email protected]>
….json / result.csv on result (#38598)
…y passed (#38599) --------- Signed-off-by: vitsai <[email protected]> Signed-off-by: vitsai <[email protected]>
so that we can optionally skip the logs/wheels/jars uploading Signed-off-by: Lonnie Liu <[email protected]>
…y support (#38529) - Add back Aliyun cluster launcher documentation that was removed (not sure why, I'm guessing it got lost in the shuffle when the cluster docs were reorganized last year.) - Group cluster launcher tabs into "Ray Team Supported" and "Community Supported" - Add vSphere to lists of community-supported cluster launchers This makes the type of support more clear to users. This is what the tabs look like: <img width="848" alt="Screenshot 2023-08-16 at 3 45 34 PM" src="https://github.com/ray-project/ray/assets/5459654/1a0e1d65-56b3-44cd-86df-5ba2a05f18f5"> <img width="772" alt="Screenshot 2023-08-16 at 3 46 06 PM" src="https://github.com/ray-project/ray/assets/5459654/4b2ee596-d974-4282-930b-a55fe4ead91f">
Previously, we define these `MapTransformFn`s as partial functions. These PR converts them to callable classes. By doing this, we can still access the original attributes, e.g., batch_size, batch_format, etc. --------- Signed-off-by: Hao Chen <[email protected]>
remove sleep in the unit test.
Signed-off-by: Sihan Wang <[email protected]>
We used to illustrate examples with Python files and the literalinclude directive. Recently, we re-wrote examples with the testcode directive. As a result, most Python files under doc_code/ are now obsolete. This PR cleans them up. Signed-off-by: Balaji Veeramani <[email protected]>
Our implementation assumes that datasources expose a positional TaskContext parameter, but it's not part of the defined interface. This PR fixes the issue by adding a ctx parameter to the defined interface. Signed-off-by: Balaji Veeramani <[email protected]>
Remove mlflow_fashion_mnist_example.py. Signed-off-by: xwjiang2010 <[email protected]>
only upload logs on postmerge, so that the tests can run on premerge when not triggered by PR. other pipelines do not have the permissions to upload. Signed-off-by: Lonnie Liu <[email protected]>
Signed-off-by: woshiyyya <[email protected]> Signed-off-by: Yunxuan Xiao <[email protected]> Co-authored-by: matthewdeng <[email protected]>
…moke tests (#38651) Signed-off-by: Justin Yu <[email protected]>
…reporting incorrect (#38685) --------- Signed-off-by: rickyyx <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
See Commits and Changes for more details.
Created by pull[bot]
Can you help keep this open source service alive? 💖 Please sponsor : )