Deployment Pipelines - Creation of New Models

[0_0]

Hey,

When there are new models created in Development/Test Workspaces, I am unable to create deployments rules for these items, how do you go about mapping the DirectLake Model to the corresponding Lakehouse in the other workspace, likewise for composite models?

Thanks

[Anonymous · Accepted solution]

Hi @0_0,

Thank you for reaching out to the Microsoft Fabric Forum Community, and special thanks to @Vinodh247  for prompt and helpful responses.

Deployment rules only work after a semantic model already exists in the target stage and has been matched by the pipeline. When you create a new DirectLake or composite model in Dev/Test, rules won’t be available at first. You must deploy the model E 2 E once without rules so Fabric can create the corresponding model in Test/Prod and establish its identity. During this first deployment, the model will still point to the Dev/Test Lakehouse or datasets, and the rule option may appear but stay disabled for a short time. After the deployment fully completes and Fabric finishes syncing (usually a few minutes, sometimes longer), the rule becomes enabled and you can remap the DirectLake connection to the correct Lakehouse in each workspace. The same limitation applies to composite models, and all referenced datasets and Lakehouses must already exist in the target workspace before rules can be configured.

https://learn.microsoft.com/en-us/fabric/cicd/deployment-pipelines/create-rules?tabs=new-ui

Thanks & Regards,

Prasanna Kumar

[Anonymous]

Hi @0_0,

Just following up to see if the Response provided by community members were helpful in addressing the issue. if the issue still persists Feel free to reach out if you need any further clarification or assistance.

Best regards,
Prasanna Kumar

[Anonymous]

Hi @0_0,

Just following up to see if the Response provided by community members were helpful in addressing the issue. if the issue still persists Feel free to reach out if you need any further clarification or assistance.

Best regards,
Prasanna Kumar

[Anonymous · Accepted solution]

Hi @0_0,

Thank you for reaching out to the Microsoft Fabric Forum Community, and special thanks to @Vinodh247  for prompt and helpful responses.

Deployment rules only work after a semantic model already exists in the target stage and has been matched by the pipeline. When you create a new DirectLake or composite model in Dev/Test, rules won’t be available at first. You must deploy the model E 2 E once without rules so Fabric can create the corresponding model in Test/Prod and establish its identity. During this first deployment, the model will still point to the Dev/Test Lakehouse or datasets, and the rule option may appear but stay disabled for a short time. After the deployment fully completes and Fabric finishes syncing (usually a few minutes, sometimes longer), the rule becomes enabled and you can remap the DirectLake connection to the correct Lakehouse in each workspace. The same limitation applies to composite models, and all referenced datasets and Lakehouses must already exist in the target workspace before rules can be configured.

https://learn.microsoft.com/en-us/fabric/cicd/deployment-pipelines/create-rules?tabs=new-ui

Thanks & Regards,

Prasanna Kumar

[Vinodh247]

This is a current limitation of Fabric deployment pipelines. When a new semantic model (DirectLake/composite) is created in Dev/Test deployment rules do not automatically appear until the model has been deployed once and exists in the target workspace. Deployment rules only work on matched items across stages, not on newly introduced ones. The practical approach is to first deploy the model end to end without rules & let fabric create the corresponding model in test & Prod, and then configure deployment rules to remap the DirectLake model to the correct lakehouse in each workspace.

For composite models, the same applies: all referenced datasets and Lakehouses must already exist in the target workspace before rules can be defined.

In short, initial deployment establishes identity; rules are only for subsequent remapping, not first time model creation.

[0_0]

Thanks for the reponse!

If we deploy a DirectLake model how long can we expect to see the rules available for the model, since if we deploy it as is it would still be mapped to the DirectLake referenced in Dev/Test workspaces. I've tried deploying as is, and once deployed, I did see an option to create a rule, but it was unavailable.

Questions:

• Would it still allow for me to configure a rule despite the source being the DirectLake in the Dev/Test Workspaces, does this also apply to composite models?
• On average how long would it take for me to set a rule upon deploying an item to higher stages?
• Can you include some documentation or highlight any notable limitations?

Thanks!

[Vinodh247 · Accepted solution]

for DirectLake semantic models, deployment rules only become configurable after Fabric successfully resolves and matches the dependent artifacts (LH, WH) in the target workspace. When you deploy “as is” the model lands in Test or Prod still pointing to the Dev/Test Lakehouse reference. Until Fabric detects a valid, compatible Lakehouse in the target stage, the data source rules stay disabled. This is not a timing delay issue in most cases; it is a dependency resolution limitation.

DirectLake and composite models behaviour?
Yes, the same behaviour applies to composite models. Rules can only be configured if all referenced sources already exist in the target workspace and are of the expected type.

• DirectLake: target Lakehouse must already exist and be recognised as a match.

• Composite: all DirectLake sources, warehouses, and imported datasets must exist. One unresolved dependency blocks rule configuration for the entire model.

How long does it take for rules to become available?
There is no guaranteed time window.

In practice:

• If dependencies exist and match, rules usually become available within a few minutes.

• If dependencies do not exist or were created after deployment, rules may remain unavailable indefinitely until you redeploy or manually align artifacts. 

This is why “waiting longer” often does not help.