refactor(sglang): split remote backend from sidecar launcher#11634
Closed
connorcarpenter15 wants to merge 1 commit into
Closed
refactor(sglang): split remote backend from sidecar launcher#11634connorcarpenter15 wants to merge 1 commit into
connorcarpenter15 wants to merge 1 commit into
Conversation
Signed-off-by: Connor Carpenter <connorc@nvidia.com>
Contributor
This comment has been minimized.
This comment has been minimized.
Member
Author
|
Superseded by #11636, which uses the internal |
This file contains hidden or 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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
dynamo-sglang-remotedynamo_sglang_remoteSglangRemoteEnginedynamo._core.backend._run_sglang_remote(argv)dynamo.sglang.sidecar.main(argv)as the actual SGLang-managed sidecar entrypointWhy
Follow-up to #11408.
The native gRPC adapter is useful as a standalone remote backend and only becomes a sidecar when SGLang manages it together with Dynamo's unified worker lifecycle. Calling the adapter itself a sidecar blurred those two responsibilities, especially for SGLang's
--sidecarmodule contract.This change keeps transport, discovery, request conversion, health gating, cancellation, and disaggregation in the remote-backend crate. The small Python module owns the SGLang sidecar composition.
Interfaces
Standalone remote backend:
SGLang-managed sidecar:
python3 -m sglang.launch_server \ <args> \ --grpc-port 30001 \ --sidecar dynamo.sglang.sidecarThe native launcher prepends a stable
argv[0], releases the GIL, and runs the shareddynamo_backend_common::runworker lifecycle.Validation
cargo test -p dynamo-sglang-remote— 21 passedcargo test --manifest-path lib/bindings/python/Cargo.toml sglang_remote_argv_prepends_program_name --libcargo fmt --all -- --checkcargo fmt --manifest-path lib/bindings/python/Cargo.toml -- --checkbash -nfor both renamed launch scriptsGPU end-to-end serving and released-wheel installation are left to CI/integration environments.