Hi Everyone
In my company we wanted to test the OAS DV capability to run Python code for ML models in Data Flows as a proof of concept in order to offer the capability to clients. To do this I followed the steps in the blogpost here https://www.rittmanmead.com/blog/2022/11/leveraging-custom-python-scripts-in-oas/ .
When testing I got the error
nQSError: 46175] Failed to download data from XSA Server: HTTP Error Code: 500. Invalid URL: http://oas.sub06191440320.datasciencevcn.oraclevcn.com:9505/datasetsvc.
[nQSError: 43224] The Dataflow "" failed during the execution.
[nQSError: 43204] Asynchronous Job Manager failed to execute the asynchronous job.
At first I thought there might be a problem with my script, so I decided to test the official example available at https://www.oracle.com/business-analytics/data-visualization/examples/ .
After I imported the example and tested the Data Flow I got a similar error
[nQSError: 46164] HTTP Server returned 404 (Not Found) for URL [masked_url].
[nQSError: 46281] Failed to download metadata for dataset 'weblogic'.'OAC.ML.SupportVectorRegression.apply'.
[nQSError: 43224] The Dataflow "SVR-AirQuality-RHPred-ApplyModel" failed during the execution.
[nQSError: 43204] Asynchronous Job Manager failed to execute the asynchronous job.
As the example did not work I examined the logs, specifically the obis1-diagnostic.log file and I saw that there are more errors related to reaching the http://oas.sub06191440320.datasciencevcn.oraclevcn.com:9505 address such as
[2024-05-13T12:31:27.802+00:00] [OBIS] [NOTIFICATION:1] [] [] [ecid: 00665Iqq4yIFKA05zzc9yW0004fQ00003T,0] [sik: ssi] [tid: 5c84d700] [nQSError: 46164] HTTP Server returned 404 (Not Found) for URL http://oas.sub06191440320.datasciencevcn.oraclevcn.com:9505/bi-servicelcm-rest/propertystore/v1/serviceinstances/ssi/systemreferences/oa.servicelogging.customer.diagnostic. [[^M
file: server/Utility/Generic/Src/SUServerLog.cpp; line: 767
This seems to be an error related to the bi_internal_channel1 from what I could see in the Weblogic console.
An additional step I did was to test the machine learning models from the OOTB installation by registering a model from OML in our ADW and I successfully ran a dataflow which uses it which means that the machine learning part should work properly.
I use OAS 2024 which runs on an Oracle Linux 7.9 using the Oracle Database 19c as the database. The OAS and Database are deployed on OCI as instances which were installed from Oracle Marketplace. I also use a Windows Server instance in the same VCN to connect to it via remote desktop and then use OAS from there.
I was wondering what could be a solution for this issue.
Regards,
Grega
