I'm trying to deploy the customer model.
I finally fond the sample templates for custom environments.
I downloaded docker drop-in environment sample files and made zip file.
[docker drop-in environment sample files]
But the upload the zip file didn't work at all.
The error message in the log tab was "network image-builder not found".
What does it means?
How can I fix it?
I did not upload any custom model file.
I just create a new custom model like the link you sent : https://docs.datarobot.com/en/docs/mlops/deployment/custom-models/custom-inf-model.html
1. Add new model on Model Registry>Custom model workshop>Models
2. Fill in some edit boxes on Add custom inference model page.
3. Click 'Add custom model'
4. the message "No environments with successful builds found" poped up
I'm not sure that this problem came from k8s installment environment.
I heard that DataRobot basically has a standalone k8s, so it means that it is no matter what is our on-prem installment environment.
Hi @cookie_yamyam ,
In order to replicate the error on our end, can you please share some details?
1. Which default environment did you select for the custom model? You need not upload any files to select some of the preloaded environments.
2. Which python libraries are used in your custom model?
Also, you can use the step-by-step process mentioned in this link to create custom models.
I think there is a problem with the installation environment.
I found the lines on docs below.
Our DataRobot is not installed in K8s, but in Docker.
That's why the error came up on drop-in.
Does it make sense?
No, maybe it didn't work.
If I register custom model in workshop, I can see this message.
I know that there is a default environment, so I don't have to drop in my custom environment.
But it seems the default environment doesn't work.
I see the error you are sharing, but may I ask you to tell me the initial problem you are trying to solve and why default environments didn't work for you?