feat: Add GRPC error codes to GRPC streaming if enabled by user. #7499
Conversation
rmccorm4
changed the title
| @@ -0,0 +1,52 @@ | |||
| # Copyright (c) 2024, NVIDIA CORPORATION. All rights reserved. | |||
There was a problem hiding this comment.
Is it possible to move the model to the L0_backend_python/lifecycle folder? I think it might be easier this way to track which test the models belong to.
There was a problem hiding this comment.
We don't have any models or model folder in
https://github.com/triton-inference-server/server/tree/main/qa/L0_backend_python/lifecycle
In test.sh we copy over from python_models and create the models folder with versions inside.
I have kept this new model in parallel to existing models used in L0_backend_python/lifecycle
There was a problem hiding this comment.
you can always create a models subfolder under L0_* tests
There was a problem hiding this comment.
thanks Olga, that's what I meant.
There was a problem hiding this comment.
In the existing test.sh we do
rm -fr *.log ./models before we start the test.
For me to add models to
L0_backend_python/lifecycle I will need to remove this.We cleanup the models before every test, I will need to remove all the instances where we delete the models folder.
server/qa/L0_backend_python/lifecycle/test.sh
Line 103 in 5320009
server/qa/L0_backend_python/lifecycle/test.sh
Line 170 in 5320009
Not sure if I should change existing design might impact other tests too.
There was a problem hiding this comment.
name the folder differently and copy from it to models ?
There was a problem hiding this comment.
@tanmayv25 suggesnted I move all the changes from L0_lifecycle to a new L0_ dedicated for this feature going forward.
The reason I made changes here was, L0_lifecycle/models had models where we send errors pragmatically helped me reuse all the models.
Will resolve this comment along with the original comment by @tanmayv25 in a new PR after the cherrypick.
Keeping it unresolved for now will mark this resolved after the new PR
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
…) (#7555) Co-authored-by: Indrajit Bhosale <iamindrajitb@gmail.com>
What does the PR do?
Enhanced GRPC streaming to enable sending GRPC error codes if enabled by user.
User can enabled/disable this feature by including "grpc_strict":true as part of GRPC headers.
Checklist
<commit_type>: <Title>Commit Type:
Check the conventional commit type
box here and add the label to the github PR.
Related PRs:
None
Where should the reviewer start?
ExtractStateFromHeaders() function and Process() function in stream_infer_handler.cc
Test plan:
Background
Customer request to have GRPC error codes as part of GRPC response in case of errors from core
TEP link : https://docs.google.com/document/d/1TfNAMYLsPuLrduBtAKo64YqV55IWQhcn6zayc_CBY18/edit?pli=1