[ray] fix: make spawn worker group hold strong reference to actors by wuxibin89 · Pull Request #1443 · verl-project/verl

wuxibin89 · 2025-05-08T13:38:17Z

Checklist Before Starting

Search for similar PR(s).

What does this PR do?

Spawned RayWorkerGroup get actors by name, which holds a weak reference to the actor and causes actors garbage collected unexpectedly. Pass actor handle explicitly in spawn to make RayWorkerGroup have strong reference to these actors. close #1365 #1138 (comment)

High-Level Design

Demonstrate the high-level design if this PR is complex.

Specific Changes

List the specific changes.

API

Demonstrate how the API changes if any.

Usage Example

Provide usage example(s) for easier usage.

# Add code snippet or script demonstrating how to use this

Test

For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluatuion results, etc.

Additional Info.

Issue Number: Fixes issue # or discussion # if any.
Training: [Note which backend this PR will affect: FSDP, Megatron, both, or none]
Inference: [Note which backend this PR will affect: vLLM, SGLang, both, or none]

Checklist Before Submitting

Read the Contribute Guide.
Apply pre-commit checks.
Add [BREAKING] to the PR title if it breaks any API.
Update the documentation about your changes in the docs.
Add CI test(s) if neccessary.

…refix=None)` (#1838) ### Checklist Before Starting - [x] Search for similar PR(s). ### What does this PR do? In #1443, a new argument `name_prefix` was introduced for function `from_detached` without setting a default value. This PR sets its default value as `None`, in which case, `RayWorkerGroup` will generate a random string as the prefix. This fix makes the API compatible with existing usage, and the users don't need to worry about this new args when a `name_prefix` is not not context necessary. ### Checklist Before Submitting - [x] Read the [Contribute Guide](https://github.com/volcengine/verl?tab=readme-ov-file#contribution-guide). - [x] Apply [pre-commit checks](https://github.com/volcengine/verl?tab=readme-ov-file#code-linting-and-formatting). - [ ] Add `[BREAKING]` to the PR title if it breaks any API. - [ ] Update the documentation about your changes in the [docs](https://github.com/volcengine/verl/tree/main/docs). - [ ] New CI unit test(s) are added to cover the code path. - [x] Rely on existing unit tests on CI that covers the code path. Signed-off-by: Hongpeng Guo <hg5@illinois.edu>

…refix=None)` (verl-project#1838) ### Checklist Before Starting - [x] Search for similar PR(s). ### What does this PR do? In verl-project#1443, a new argument `name_prefix` was introduced for function `from_detached` without setting a default value. This PR sets its default value as `None`, in which case, `RayWorkerGroup` will generate a random string as the prefix. This fix makes the API compatible with existing usage, and the users don't need to worry about this new args when a `name_prefix` is not not context necessary. ### Checklist Before Submitting - [x] Read the [Contribute Guide](https://github.com/volcengine/verl?tab=readme-ov-file#contribution-guide). - [x] Apply [pre-commit checks](https://github.com/volcengine/verl?tab=readme-ov-file#code-linting-and-formatting). - [ ] Add `[BREAKING]` to the PR title if it breaks any API. - [ ] Update the documentation about your changes in the [docs](https://github.com/volcengine/verl/tree/main/docs). - [ ] New CI unit test(s) are added to cover the code path. - [x] Rely on existing unit tests on CI that covers the code path. Signed-off-by: Hongpeng Guo <hg5@illinois.edu>

…erl-project#1443) ### Checklist Before Starting - [ ] Search for similar PR(s). ### What does this PR do? Spawned RayWorkerGroup get actors by name, which holds a weak reference to the actor and causes actors garbage collected unexpectedly. Pass actor handle explicitly in spawn to make RayWorkerGroup have strong reference to these actors. close verl-project#1365 verl-project#1138 (comment) ### High-Level Design > Demonstrate the high-level design if this PR is complex. ### Specific Changes > List the specific changes. ### API > Demonstrate how the API changes if any. ### Usage Example > Provide usage example(s) for easier usage. ```python # Add code snippet or script demonstrating how to use this ``` ### Test > For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluatuion results, etc. ### Additional Info. - **Issue Number**: Fixes issue # or discussion # if any. - **Training**: [Note which backend this PR will affect: FSDP, Megatron, both, or none] - **Inference**: [Note which backend this PR will affect: vLLM, SGLang, both, or none] ### Checklist Before Submitting - [ ] Read the [Contribute Guide](https://github.com/volcengine/verl?tab=readme-ov-file#contribution-guide). - [ ] Apply [pre-commit checks](https://github.com/volcengine/verl?tab=readme-ov-file#code-linting-and-formatting). - [ ] Add `[BREAKING]` to the PR title if it breaks any API. - [ ] Update the documentation about your changes in the [docs](https://github.com/volcengine/verl/tree/main/docs). - [ ] Add CI test(s) if neccessary.

…refix=None)` (verl-project#1838) ### Checklist Before Starting - [x] Search for similar PR(s). ### What does this PR do? In verl-project#1443, a new argument `name_prefix` was introduced for function `from_detached` without setting a default value. This PR sets its default value as `None`, in which case, `RayWorkerGroup` will generate a random string as the prefix. This fix makes the API compatible with existing usage, and the users don't need to worry about this new args when a `name_prefix` is not not context necessary. ### Checklist Before Submitting - [x] Read the [Contribute Guide](https://github.com/volcengine/verl?tab=readme-ov-file#contribution-guide). - [x] Apply [pre-commit checks](https://github.com/volcengine/verl?tab=readme-ov-file#code-linting-and-formatting). - [ ] Add `[BREAKING]` to the PR title if it breaks any API. - [ ] Update the documentation about your changes in the [docs](https://github.com/volcengine/verl/tree/main/docs). - [ ] New CI unit test(s) are added to cover the code path. - [x] Rely on existing unit tests on CI that covers the code path. Signed-off-by: Hongpeng Guo <hg5@illinois.edu>

…erl-project#1443) ### Checklist Before Starting - [ ] Search for similar PR(s). ### What does this PR do? Spawned RayWorkerGroup get actors by name, which holds a weak reference to the actor and causes actors garbage collected unexpectedly. Pass actor handle explicitly in spawn to make RayWorkerGroup have strong reference to these actors. close verl-project#1365 verl-project#1138 (comment) ### High-Level Design > Demonstrate the high-level design if this PR is complex. ### Specific Changes > List the specific changes. ### API > Demonstrate how the API changes if any. ### Usage Example > Provide usage example(s) for easier usage. ```python # Add code snippet or script demonstrating how to use this ``` ### Test > For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluatuion results, etc. ### Additional Info. - **Issue Number**: Fixes issue # or discussion # if any. - **Training**: [Note which backend this PR will affect: FSDP, Megatron, both, or none] - **Inference**: [Note which backend this PR will affect: vLLM, SGLang, both, or none] ### Checklist Before Submitting - [ ] Read the [Contribute Guide](https://github.com/volcengine/verl?tab=readme-ov-file#contribution-guide). - [ ] Apply [pre-commit checks](https://github.com/volcengine/verl?tab=readme-ov-file#code-linting-and-formatting). - [ ] Add `[BREAKING]` to the PR title if it breaks any API. - [ ] Update the documentation about your changes in the [docs](https://github.com/volcengine/verl/tree/main/docs). - [ ] Add CI test(s) if neccessary.

…refix=None)` (verl-project#1838) ### Checklist Before Starting - [x] Search for similar PR(s). ### What does this PR do? In verl-project#1443, a new argument `name_prefix` was introduced for function `from_detached` without setting a default value. This PR sets its default value as `None`, in which case, `RayWorkerGroup` will generate a random string as the prefix. This fix makes the API compatible with existing usage, and the users don't need to worry about this new args when a `name_prefix` is not not context necessary. ### Checklist Before Submitting - [x] Read the [Contribute Guide](https://github.com/volcengine/verl?tab=readme-ov-file#contribution-guide). - [x] Apply [pre-commit checks](https://github.com/volcengine/verl?tab=readme-ov-file#code-linting-and-formatting). - [ ] Add `[BREAKING]` to the PR title if it breaks any API. - [ ] Update the documentation about your changes in the [docs](https://github.com/volcengine/verl/tree/main/docs). - [ ] New CI unit test(s) are added to cover the code path. - [x] Rely on existing unit tests on CI that covers the code path. Signed-off-by: Hongpeng Guo <hg5@illinois.edu>

…erl-project#1443) ### Checklist Before Starting - [ ] Search for similar PR(s). ### What does this PR do? Spawned RayWorkerGroup get actors by name, which holds a weak reference to the actor and causes actors garbage collected unexpectedly. Pass actor handle explicitly in spawn to make RayWorkerGroup have strong reference to these actors. close verl-project#1365 verl-project#1138 (comment) ### High-Level Design > Demonstrate the high-level design if this PR is complex. ### Specific Changes > List the specific changes. ### API > Demonstrate how the API changes if any. ### Usage Example > Provide usage example(s) for easier usage. ```python # Add code snippet or script demonstrating how to use this ``` ### Test > For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluatuion results, etc. ### Additional Info. - **Issue Number**: Fixes issue # or discussion # if any. - **Training**: [Note which backend this PR will affect: FSDP, Megatron, both, or none] - **Inference**: [Note which backend this PR will affect: vLLM, SGLang, both, or none] ### Checklist Before Submitting - [ ] Read the [Contribute Guide](https://github.com/volcengine/verl?tab=readme-ov-file#contribution-guide). - [ ] Apply [pre-commit checks](https://github.com/volcengine/verl?tab=readme-ov-file#code-linting-and-formatting). - [ ] Add `[BREAKING]` to the PR title if it breaks any API. - [ ] Update the documentation about your changes in the [docs](https://github.com/volcengine/verl/tree/main/docs). - [ ] Add CI test(s) if neccessary.

…refix=None)` (verl-project#1838) ### Checklist Before Starting - [x] Search for similar PR(s). ### What does this PR do? In verl-project#1443, a new argument `name_prefix` was introduced for function `from_detached` without setting a default value. This PR sets its default value as `None`, in which case, `RayWorkerGroup` will generate a random string as the prefix. This fix makes the API compatible with existing usage, and the users don't need to worry about this new args when a `name_prefix` is not not context necessary. ### Checklist Before Submitting - [x] Read the [Contribute Guide](https://github.com/volcengine/verl?tab=readme-ov-file#contribution-guide). - [x] Apply [pre-commit checks](https://github.com/volcengine/verl?tab=readme-ov-file#code-linting-and-formatting). - [ ] Add `[BREAKING]` to the PR title if it breaks any API. - [ ] Update the documentation about your changes in the [docs](https://github.com/volcengine/verl/tree/main/docs). - [ ] New CI unit test(s) are added to cover the code path. - [x] Rely on existing unit tests on CI that covers the code path. Signed-off-by: Hongpeng Guo <hg5@illinois.edu>

[ray] fix: make spawn worker group hold strong reference to actors

0247390

wuxibin89 requested review from PeterSH6, eric-haibin-lin and vermouth1992 May 8, 2025 13:38

wuxibin89 mentioned this pull request May 8, 2025

[rollout] feat: introduce vLLM AsyncLLM to support multi-turn rollout #1138

Merged

5 tasks

vermouth1992 approved these changes May 8, 2025

View reviewed changes

wuxibin89 merged commit f90b717 into main May 8, 2025
32 of 33 checks passed

wuxibin89 deleted the wuxibin/fix_wg_ref branch May 8, 2025 15:08

ruiyiw mentioned this pull request May 12, 2025

Multi-Turn Rollout on a single GPU #1365

Closed

hongpeng-guo mentioned this pull request Jun 4, 2025

[fix] Adding a default value for RayWorkerGroup.from_detached(name_prefix=None) #1838

Merged

7 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[ray] fix: make spawn worker group hold strong reference to actors#1443

[ray] fix: make spawn worker group hold strong reference to actors#1443
wuxibin89 merged 1 commit intomainfrom
wuxibin/fix_wg_ref

wuxibin89 commented May 8, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

wuxibin89 commented May 8, 2025

Checklist Before Starting

What does this PR do?

High-Level Design

Specific Changes

API

Usage Example

Test

Additional Info.

Checklist Before Submitting

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants