Skip to content

Fail requests on worker error - #200

Open
NSagan271 wants to merge 4 commits into
mainfrom
fix/fail-requests-on-worker-error
Open

Fail requests on worker error#200
NSagan271 wants to merge 4 commits into
mainfrom
fix/fail-requests-on-worker-error

Conversation

@NSagan271

Copy link
Copy Markdown
Collaborator

What does this PR do?

Closes #123. When errors raise in the worker or engine, fail the corresponding requests/batch (stopping the request sending a 500 back to the client) instead of raising within the worker and leading to an infinite "Request Failed" loop. Wherever possible, fails individual requests instead of the full batch (e.g., prepare_inputs, postprocess, check_stop)

How was it tested?

New unit test.

Injected failures into orpheus prepare_inputs, preprocess, and check_stop and verified intended behavior (errors propagated to the benchmark client, unaffected requests produce full output).

Checklist

  • ruff check . passes
  • Added or updated tests / docs where relevant

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Propagate runtime inference errors back to the client

1 participant