Since the deprecation of the legacy macro‑based Tableau Update tool and the switch to the Python‑based Tableau Output Tool, we’ve been seeing extremely concerning and inconsistent behavior. In short: workflows explicitly configured to publish only to Production are intermittently updating QA instead, with no QA‑targeted tool present anywhere in the workflow.
This strongly suggests either server context leakage, DCM credential crossover, or a regression introduced since the macro tool was fully sunset, none of which support has been able to explain.
Since the deprecation of the legacy macro‑based Tableau Update tool and the move to the Python‑based Tableau Output Tool, we’ve been experiencing unsafe and inconsistent behavior when publishing Hyper extracts to Tableau Server. This is occurring with Tableau_py3.10_v1.5.4.
Specifically, workflows configured to publish only to the Production Tableau Server are intermittently publishing to QA instead, with no QA‑targeted output tool anywhere in the workflow.
-Single Tableau Output Tool per workflow
-Explicitly configured for Production
-No QA server references in the workflow
Despite this:
QA data sources are being created or replaced
Production data sources are sometimes not updated at all
In one test, I manually deleted the QA data source, ran a Prod‑only workflow, and watched the QA data source get recreated while Prod was untouched.
To head off common suggestions, this is not:
-Multiple Tableau Output tools
-Dual‑server publishing (we removed this design entirely)
-Project name collisions
-Permissions issues
-A workflow cache issue
Suspected Area: DCM
While I can’t definitively prove root cause, the behavior strongly suggests DCM credential or server context crossover, or at least I thought it did because I could sync the DCM Desktop connections to the server and it would resolve the issue sometimes; However, even this has not been working lately.
We opened a formal support ticket, collected and submitted logs with our Alteryx Admin, and the ticket was closed without any findings or explanation. At this point, I’m looking for community or engineering insight, because publishing to the wrong Tableau Server is a serious risk in an enterprise environment.
Has anyone else seen the Tableau Output Tool publish to the wrong server?
Any insight into how server context is resolved with DCM?
Known bugs, regressions, or reliable mitigations?
Appreciate any guidance, this issue has been ongoing since the macro tool was fully sunset, and we’re trying to get it to a real resolution.
