I have Splunk Free installed on AWS using the Splunk AMI (AWS linux). The docker environment is running on the same AWS instance where splunk is installed.
When I try to run the pytorch iris model, I get an an error message that the Endpoint URL is not HTTPS.
Search command:
| inputlookup iris.csv | sample partitions=10 seed=42 | where partition_number<7 | fit MLTKContainer algo=pytorch_nn epochs=10 feature_variables=sepal_length, sepal_width, petal_length, petal_width target_variables=species into app:PyTorch_iris_model_nn
Errors:
My setup:
Splunk DLTK 3.5, MLTK 5.2.1, Python for Scientific Computing 2.0.2
Docker host is unix://var/run/docker.sock
Endpoint URL: localhost
External URL: set to public IP address of AWS Instance
When I click the Test&Save button on setup, the popup says 'Successfullly etablished connection to the container environment' and I can start the dev container and access Jupyter Notebook, MLFlow, and Tensorboard.
I tried the external endpoint url to start with https, but get the same error.
Hello friend,
I have experienced the same issues and would like to share how to resolve them.
I carefully executed the following steps precisely in that order to make it work.
1. Add the splunk user to docker group on host machine
$ sudo usermod -aG docker splunk
2. Create self-signed certificates and create the server certificates https://docs.splunk.com/Documentation/Splunk/8.2.1/Security/Howtoself-signcertificates
3. Create a single PEM file https://docs.splunk.com/Documentation/Splunk/8.2.1/Security/HowtoprepareyoursignedcertificatesforSpl...
4. Configure indexers to use a signed certificates https://docs.splunk.com/Documentation/Splunk/8.2.1/Security/ConfigureSplunkforwardingtousesignedcert...
After applying those configurations to our single-instance the DLTK Endpoint URL() is not HTTPS error was resolved.
Hope that helps 😉