hydrospheredata / hydro-serving-manager Goto Github PK

Due to the poor integration with spotify-docker, docker push fails silently and we can't handle errors.

Seems like the could be handled only in ProgressHandler which is passed to the push method.

• maxLifetime
• leakDetectionThreshold

If deployment configuration is incorrect, k8s will return 400 with error message.

Now we don't parse it, and handle it as an unexpected exception with 500 code. Error message is propagated though.

Steps to reproduce:
Give your model a name consisting with upper-case letters.

kind: Model
name: Linear_regression
runtime: "hydrosphere/serving-runtime-python-3.6:2.1.0"
install-command: "pip install -r requirements.txt"

The model will not build and the UI will show 0 in its building logs .

Docker logs:

java.lang.IllegalStateException: Could not acquire image ID or digest following build

Couldn't find a way to know which servable is used in which application.

Now we only have option to get model version by using name/version pattern.

• Check Servables status and their cloud instances
• Check if there are any ModelVersion in Assembling status. Try to restart the build, or fail instantly.

For instance in docker env trying to delete Servable with crashed container you will get Internal server error.

When creating an application without specifying deployment configurations, we receive an error:

org.postgresql.util.PSQLException: ERROR: insert or update on table "servable" violates foreign key constraint "servable_deploy ment_configuration_fkey"