Configuring the Python Repositories
Configuring pip and Conda environmental variables
You can define pip and Conda key-value pairs as:
- Environmental variables for shell and Jupyter
- Default build arguments for Nuclio and MLRun
The values are applied to every instance of these services. You can define any variable that is supported by pip or Conda. Changing the Pypi server URL causes a restart of Jupyter, MLRun, Nuclio, and Shell, services.
Configuring the variables in the Dashboard
To configure pip and Conda environmental variables for your cluster:
Open the platform dashboard and select the settings gear-wheel icon () from the top-right toolbar of any page to open the
Python packages, then under PIP Options, press Create a new optionand fill in one or more key-value pairs.
Under Conda options, press
Create a new optionand fill in one or more key-value pairs.
Apply. A loading spinner displays until the values are propagated to the services.
To delete a key-value pair, press the
Deleteicon next to the row of the key-value pair.
Creating a PyPI certificate secret and loading it to Kubernetes
Run this procedure if you have an unsigned certificate in a closed network, so that Provazio can inject the certificate into MLRun and Nuclio.
Create a secret in the relevant namespace (
default-tenant) with the name user-custom-pypi-cert and this skeleton:
apiVersion: v1 kind: Secret metadata: name: user-custom-pypi-cert data: pip-ca-certificates.crt: <CERT_CONTENT> ```
Restart MLRun and Nuclio.