We are experiencing an intermittent refresh issue with an Oracle Analytics Cloud dataset that uses a REST API (Preview) connection.
The dataset is used in a workbook join to apply user visibility/security filtering. It retrieves two fields from a Fusion REST endpoint:
UserNameSecurityContextValue
The refresh behavior is alternating:
- some executions complete successfully;
- other executions fail a few minutes later, without any configuration change;
- the history currently shows 92 successful full refreshes and 22 failed refreshes.
The dataset is configured with Automatic Caching and replaces the existing data on each full refresh.
A successful refresh reads 9 rows and completes normally. The OAC Diagnostic Log shows the following sequence:
Data Refresh Service
-> logical query against the XSA dataset object
-> BI Server queue time: 0.006 seconds
-> physical query sent through the REST connector:
select "USERNAME", "SECURITYCONTEXTVALUE" from AUTOREST.ITEMS
-> Query Status: Successful Completion
The physical query completes in approximately 0.82 seconds after it is sent to AUTOREST.ITEMS.
For a failed refresh, the Diagnostic Log shows:
Dataset reload job started
-> logical query created by Data Refresh Service
-> BI Server queue time: 0.003 seconds
-> [nQSError: 47123] Invalid connection details were supplied.
Please enter the correct details and try again.
-> [nQSError: 43119] Query Failed
-> Dataset reload task ended with status: Failed
In the failing case, the log does not show the subsequent physical-query step:
Sending query to database ... from AUTOREST.ITEMS
Therefore, the failure appears to occur after OAC resolves the logical dataset query, but before the physical query is dispatched through the REST connector.
The REST connection uses Basic Authentication with a Fusion SaaS technical account. The connection configuration, endpoint, and credentials have not been changed between successful and failed executions.
Could anyone advise on:
- What conditions can cause intermittent
nQSError: 47123 Invalid connection details were supplied for an OAC REST API dataset? - Does this error indicate a connection/credential resolution issue within OAC, or can it also be returned when the remote Fusion endpoint rejects a request?
- Are there known limitations involving connection pooling, credential rotation, throttling, concurrency, or REST API usage limits that could cause this alternating behavior?
- Which additional OAC diagnostic logs or settings would help capture the underlying HTTP status or connector error?