Creation of services in OSCAR #7
Conversation
vicente87
commented
Jul 13, 2023
vicente87
commented
- The codes of the services are modified to adapt them to the asynchronous invocations of OSCAR.
- A new service (EMG) is created that is used to join the N embedding and pass them as input parameters to the DDS.
- Several elements are created in the Makefile to create the images of the services.
- The creation_service folder has the FDL files to create the services in the different OSCAR clusters
creation_service/ML_inference/all.sh
Outdated
There was a problem hiding this comment.
@vicente87 This file is empty.
There was a problem hiding this comment.
@jaime-cespedes-sisniega You are right, this file does not have to be there.
There was a problem hiding this comment.
@vicente87 This file is empty.
There was a problem hiding this comment.
@jaime-cespedes-sisniega You are right, this file does not have to be there.
There was a problem hiding this comment.
@vicente87 This file is empty.
There was a problem hiding this comment.
@jaime-cespedes-sisniega You are right, this file does not have to be there.
There was a problem hiding this comment.
@vicente87 This file is empty.
There was a problem hiding this comment.
@jaime-cespedes-sisniega You are right, this file does not have to be there.
There was a problem hiding this comment.
@vicente87 This file must not be modified. It is generated by the training.py script.
There was a problem hiding this comment.
@vicente87 Probably most of these packages are not required for this service.
There was a problem hiding this comment.
@jaime-cespedes-sisniega Here you are absolutely right. What happened was that the service creation pattern that you provided us was used and when we created the EMG service, we forgot to remove the libs that were not going to be used.
There was a problem hiding this comment.
@vicente87 This file must not be modified. It is generated by the training.py script.
There was a problem hiding this comment.
@vicente87 Why do you create this main-service.py when there is a main.py?
There was a problem hiding this comment.
@jaime-cespedes-sisniega The main-service.py is the modification of the main.py created by you, where the functions created have been reused to accommodate them to the invocation of MinIO services. Here output files are created so that they can be used by subsequent services and then conform to the functions created by you. But basically what the main-service.py (of the MLIS, DRS and DDS services) have is to accommodate the inputs and outputs of the files left in the MinIO buckets to be used by the previously created functions. Which perhaps could have been called main.py to follow the same structure that you created.
| @@ -28,7 +29,9 @@ build-push-dimensionality-reduction: | |||
| docker buildx build --platform linux/amd64,linux/arm64 -t ifcacomputing/dimensionality-reduction-api --push dimensionality_reduction_api | |||
|
|
|||
| build-push-model-inference: | |||
| docker buildx build --platform linux/amd64,linux/arm64 -t ifcacomputing/model-inference-api:latest --push model_inference_api | |||
| docker build -t ghcr.io/grycap/mls-arm-api model_inference_api | |||
There was a problem hiding this comment.
@vicente87 Why do you change the registry where this image is stored?
| docker buildx build --platform linux/amd64,linux/arm64 -t ghcr.io/grycap/dds-arm-api --push detector_api | ||
|
|
||
| emc: |
There was a problem hiding this comment.
@vicente87 I think that this should be the unique "new" service. The addition of the others (mls, dds and drs) is completely redundant. Because they are already hosted in ifcacomputing/XXXX-api.
There was a problem hiding this comment.
@jaime-cespedes-sisniega The problem is that in order to make the asynchronous invocations we create main-service.py (it's from what was put in main.py) and there we introduce small changes to be able to orchestrate the services well among others. For example we create output files in the drs services in json format so that you can use it well. Also, the input data that comes from the MinIO buckets has to accommodate the functions that you have created and that is why we decided to create the new images. I don't know if I will explain well.
There was a problem hiding this comment.
@vicente87 Since there is a conflict with some objects that are generated with the ml/mnist/training.py script, I suggest deleting the ones you intend to upload, since the latest version of them is already in the repo and these that are intended to be uploaded are already old versions. Thank you.
