Details
-
Bug
-
Resolution: Unresolved
-
Normal
-
None
-
None
-
2
-
False
-
-
False
Description
Deploy type
RHODS 1.34 RC2
Version
RHODS 1.34 RC2
Environment
- RHODS 1.34 RC2
- OpenShift Pipelines 1.12.1
Current Behavior
In pr-133 ObjectStorage Health Check was introduced.
If you configure a data connection using proper data for the fields Access Key, Secret key, Endpoint and Region but using a non existing bucket name, the ObjectStoreAvailable CR condition for that DataSciencePipelinesApplication doesn't detect the problem and shows Status: True and Message: Object Store connectivity successfully verified.
Some other conditions fail in this case (see screenshot below)
incorrect-bucket-name-dspa-conditions
In case you use an incorrect access key, the ObjectStoreAvailable CR condition properly informs about the problem, as seen in this screenshot
incorrect-access-key-dspa-conditions
Expected Behavior
ObjectStoreAvailable CR condition should detect when the bucket doesn't exist
Workaround (if any)
- Log in into RHODS Dashboard
- Create Data Science Project
- Create a data connection with correct data except for bucket name, using a bucket name that doesn't exist
- "Create Pipeline Server"
- In OpenShift Console, search the DataSciencePipelineApplication created in the previous step and check the CR Conditions
- Verify the status and message for the ObjectStoreAvailable CR Condition
Anything else
When the failure provoked by wrong access key happens, even if the ObjectStoreAvailable properly detects there is something wrong connecting to the storage, RHODS Dashboard just shows that there was an error with the pipeline server. I think this will be improved here: https://github.com/opendatahub-io/odh-dashboard/issues/1224
incorrect-bucket-name-dashboard-error
Migrated from GitHub: https://github.com/opendatahub-io/data-science-pipelines-operator/issues/441