Can't edit the cluster created by mlflow model serving

[ravi121]

I'm trying to deploy a ml model into production using mlflow. while in that process, I have registered the model to mlflow models. After that it created the cluster but then it was in pending state forever. when I checked the model events, I see a problem with https proxy, we have global init scripts which contain proxy information.

but databricks mentioned that global init scripts are not run on model serving created clusters. Ref: https://docs.databricks.com/applications/mlflow/model-serving.html

so the only way for us to edit the cluster and add them but in that process we are getting an error "error: Cannot edit cluster created by ModelServing".

In the "Model Events page", I see the following logs,

[Errno 101] Network is unreachable',)': /simple/mlflow/ WARNING: Retrying (Retry(total=3, connect=None, read=None, redirect=None, status=None)) after connection broken by 'NewConnectionError('<pip._vendor.urllib3.connection.HTTPSConnection object at 0x7f258247f710>: Failed to establish a new connection:

can anyone help us what is the solution to this ?

[Kaniz_Fatma · Accepted solution]

Hi @ravi g​ , To customize the serving cluster, use the Cluster Settings tab on the Serving tab.

• To modify the memory size and number of cores of a serving cluster, use the Instance Type drop-down menu to select the desired cluster configuration. When you click Save, the existing cluster is terminated and a new cluster is created with the specified settings.
• To add a tag, type the name and value in the Add Tag fields and click Add.
• To edit or delete an existing tag, click one of the icons in the Actions column of the Tags table.

Source

[Anonymous]

Hey @ravi g​ 

Does @Kaniz Fatma​'s answer help? If it does, would you be happy to mark it as best? If it doesn't, please tell us so we can help you.

Thanks!