Dataset scheduled refresh fails with odata azure devops source

[decarsul]

Hi there!

I have a report that i made, using an odata query to the azure devops environment. This was done because i wanted board columns and other items that are not available by creating an analytics view.

Now the issue is, that it will not refresh unless i am actually logged into my workspace.

I cannot find a workaround to this. Any suggestions?

[V-yubandi-msft]

Hi @decarsul ,

May I ask if you have resolved this issue? If so, please mark the helpful reply and accept it as the solution. This will be helpful for other community members who have similar problems to solve it faster.

 

Thank you.

[V-yubandi-msft]

Hi @decarsul ,

May I ask if you have resolved this issue? If so, please mark the helpful reply and accept it as the solution. This will be helpful for other community members who have similar problems to solve it faster.

 

Thank you.

[decarsul · Accepted solution]

Well, its an answer, but if acceptable... i don't think so, hence i won't mark any as such.

[V-yubandi-msft]

Hi @decarsul ,

Thanks for the update.  You're absolutely right to expect a workaround, and while Power BI doesn’t currently support scheduled refresh for Azure DevOps OData with a service principal. If there’s a specific workaround or approach you’re looking for ? Let us know .

Regards,

Yugandhar.

[BhavinVyas3003]

Use a Power BI Dataflow instead of connecting directly in the report.

In Power BI Service, go to your workspace.

Click “New” → “Dataflow”.

Choose “Add new tables”, then select OData feed as the source.

Enter your Azure DevOps OData URL (e.g., https://analytics.dev.azure.com/{org}/_odata/v3.0-preview/).

Sign in using OAuth2 (Organizational account).

Select the tables and save the dataflow.

In Power BI Desktop, use “Power BI dataflows” as your source and connect to this data.

Publish the report. Scheduled refresh will now work from the dataflow.

[decarsul]

I'll try create a dataflow, but honestly i do not see how authentication for a dataflow would be different from a dataset. Both are tied to my personal credentials, which seem to require an active AD user.

[BhavinVyas3003]

You're right, both use your Azure AD credentials but dataflows handle authentication more reliably in the cloud. Once set up, the dataflow refreshes data independently, and your dataset connects to that refreshed data, avoiding direct login issues with Azure DevOps OData.

[decarsul]

Does not work. 

Fail message is the same as with a dataset: The credentials provided for the OData source are invalid.
Works just fine when i have an active AD session.

[V-yubandi-msft]

Hi @decarsul ,

I understand how frustrating it is to encounter the  credentials are invalid error when you're not logged in. Power BI connects to Azure DevOps using your Azure login through an OData feed. The problem arises because your login session automatically expires after a few hours, causing Power BI to lose access and resulting in scheduled refresh failures.

Unfortunately, Power BI doesn't support using a system account (like a Service Principal) for Azure DevOps OData feeds, so it depends on an active user session. Consequently, the refresh will fail unless you're logged in.

If my response solved your query, please mark it as the Accepted solution to help others find it easily. And if my answer was helpful, I'd really appreciate a 'Kudos'.

[decarsul]

Dataflow created and auto refresh set for tonight 5am. Because i don't intend to work for atleast 8 hours prior to that, this should give AD enough time to remove my active session.

Will come back tommorow with the result.