-
Notifications
You must be signed in to change notification settings - Fork 58
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
LIN-632, LIN-618 Ray Integration #862
Conversation
49dcf6c
to
65c7e77
Compare
1377ffc
to
310d0e5
Compare
994fc40
to
5399e61
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
looks good. please run the housing example with inter task communication for sanity checks and then merge.
Done, synced up and made sure housing example with inter task communication works. |
# Description Adds support for ray dags. Setting framework= RAY in to_pipeline now produces a dag file that can be run using ray. Ray DAGS currently use `.remote()` instead of the new [alpha workflow](https://docs.ray.io/en/latest/workflows/index.html) API with `.bind()`. No setup and teardown tasks are added because Ray also does not have a mechanism to specify execution order to allow them to be run in the correct places. All sink nodes in the graph are materialized since RAY does not have a "run whole dag" API Fixes # (issue) LIN-632, LIN-618 ## Type of change Please delete options that are not relevant. - [X] New feature (non-breaking change which adds functionality) # How Has This Been Tested? Snapshot and e2e tests added following airflow example including a new github action to run Ray tests.
# Description Adds new RAY workflow as the default for RAY pipelines. This change modifies the behavior described in #862 * Ray now supports `use_workflows`, setting to True (which is also default) uses the new [Ray workflows API ](https://docs.ray.io/en/latest/workflows/index.html) * Adds ability to configure the runtime_env and storage options in ray.init() * Updated tests since some workflows are not representable in new Ray workflows API Fixes LIN-738 ## Type of change Please delete options that are not relevant. - [X] New feature (non-breaking change which adds functionality) # How Has This Been Tested? Ray tests updated. Run with housing example using both "use_workflows" as True and False to ensure both new and old DAGs run. [LIN-738]: https://linea.atlassian.net/browse/LIN-738?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ
Description
Adds support for ray dags. Setting framework= RAY in to_pipeline now produces a dag file that can be run using ray.
Ray DAGS currently use
.remote()
instead of the new alpha workflow API with.bind()
.No setup and teardown tasks are added because Ray also does not have a mechanism to specify execution order to allow them to be run in the correct places.
All sink nodes in the graph are materialized since RAY does not have a "run whole dag" API
Fixes # (issue)
LIN-632, LIN-618
Type of change
Please delete options that are not relevant.
How Has This Been Tested?
Snapshot and e2e tests added following airflow example including a new github action to run Ray tests.