Dynamics Community Forum Thread Details

Automating connection reference handling during solution import, is there an official approach?

Hi everyone,

We built an automated Power Platform deployment pipeline using Azure Automation (PowerShell runbooks), Microsoft.Xrm.Data.PowerShell, and App Registration auth. The flow works: export solution → store ZIP in Azure Blob → import into target environment. It breaks as soon as flows with connection references are involved.

The error we get:

MissingDependencies — connectionreference required by Workflow

What we confirmed:

Connection references ARE visible inside the solution in make.powerapps.com

BUT when we query solutioncomponent filtered by solutionid, we get 0 results for connection references

The componenttype values we see are only: Workflow, Canvas App, and two unrecognized empty entries

# Returns 0 for componenttype = 10067
$components = Get-CrmRecords -conn $conn `
    -EntityLogicalName solutioncomponent `
    -FilterAttribute solutionid -FilterOperator eq -FilterValue $solutionId `
    -Fields objectid, componenttype

$components.CrmRecords | Where-Object { $_.componenttype -eq 10067 }

So either the componenttype for connection references is not 10067, or connection references are not included as solution components the way we expect.

What we are trying to achieve:
Fully automated deployment with zero manual steps:

Detect connection references used by flows in the solution

At import time, map or create equivalent connection references in the target environment

Associate them to existing connections owned by a service account

Import the solution and have flows ready to activate

Our open questions:

What is the correct componenttype value for connection references in solutioncomponent?

Is the deployment settings JSON (pac solution create-settings) the official mechanism for this, and does it work with Import-CrmSolutionAsync or only with PAC CLI?

Can connection references be created programmatically in the target env via Dataverse API or PowerShell, or do they always require interactive user authentication?

How should source → target mapping work when connection IDs differ across environments?

Does the behavior differ between Managed and Unmanaged solutions, or between Service Principal and user-owned connections

Context: We are not using Azure DevOps or Power Platform Pipelines. Our orchestration is Azure Automation + Power Automate, triggered from a Canvas App.

Any pointer to official documentation or recommended patterns would be very helpful.

Thanks in advance.

Categories:

Power Platform integration

Quick Links

Leaderboard > Microsoft Dynamics 365 | Integration, Dataverse, and general topics