-
Bug
-
Resolution: Unresolved
-
Normal
-
None
-
RHOAI_2.6.0_GA
-
False
-
-
False
-
-
-
Important
Deploy type
Downstream version (eg. OpenShift AI 2.6)
Version
RHOAI 2.6 Pre-RC (brew.registry.redhat.io/rh-osbs/iib:653804)
Current Behavior
Following the UI to deploy a model, there's is no instructions to initially have a model file in the bucket.
In such case, the model deployment will fail upon:
INFO:root:Initializing, args: src_uri [s3://ods-ci-s3/con1] dest_path[ [/mnt/models] INFO:root:Copying contents of s3://ods-ci-s3/con1 to local INFO:botocore.credentials:Found credentials in environment variables. Traceback (most recent call last): File "/storage-initializer/scripts/initializer-entrypoint", line 15, in <module> Storage.download(src_uri, dest_path) File "/kserve/kserve/storage/storage.py", line 83, in download Storage._download_s3(uri, out_dir) File "/kserve/kserve/storage/storage.py", line 244, in _download_s3 raise RuntimeError( RuntimeError: Failed to fetch model. No model found in con1.
Expected Behavior
The dialog was confusing, as if all we need is just to enter the connection to the Bucket:
It is confusing since the common title "Existing / New data connection" title is widely used in similar DS dialogs.
But here it should be cleared to the User, that the entered connection should be created only after the bucket already includes a model file or directory with a model.
Can you add a red comment bellow the connection dialog like:
Make sure to upload the model file to the destination path in advanced.
Steps To Reproduce
Workaround (if any)
What browsers are you seeing the problem on?
Anything else