CRM Analytics dataflow warnings and errors

Below are the specific warnings and errors that a CRM Analytics dataflow job can produce, with links to the individual articles documenting resolutions.

Warnings

Warnings mean the dataflow completed and all datasets registered successfully, but a resulting data quality issue may exist. Expanding the dataflow job in the Data Manager reveals which nodes are responsible. Administrators familiar with the dataflow structure and source data will usually need to investigate the root cause. If the warning occurs on a templated app, it could mean the app is referencing a feature not in use in this environment.

The dataflow was completed, but augment node performed a LookupMultiValue augment operation on a date field, which can give unexpected results
This message occurs when a date field is in the "right_select" of an augment transformation of a "LookupMultiValue" operation. This can produce undesired date values in the resulting datasets.

The dataflow was completed, but the node didn't augment any columns
In an augment transformation, this warning is produced when the key from the left dataset has no matches to the key on the right dataset. This means that no information from the right dataset is included in any of the rows.

The sharing source and security predicate in this dataset version must be the same as in the dataflow
This warning is received when the Sharing Source in a dataflow register node does not match those defined in the target dataset's edit page.

Errors

Errors mean the dataflow failed and none of its datasets were registered. Re-running the dataflow can overcome a transient issue. If the failure persists, the root cause will need deeper investigation by an Administrator, and it may be necessary to engage Salesforce Support.

Can't find dimension in update dataset
A key used in an augment transformation does not exist or its values are entirely null.

Can't propagate sharing rules through the augment node
The left and right sources in an augment transformation contain an sfdcDigest of an object in their lineages, and then that object is used as a Sharing Source for a register node.

Error accessing replicated dataset.: Invalid extract field name
In environments with Data Sync enabled, an sfdcDigest node contains a field that is not present in the Connected Dataset.

Dataflow Error: Error during local fetch: Replicated dataset was not found
In environments with Data Sync enabled, an sfdcDigest node references an object with no corresponding Connected Dataset. This can also occur during an org migration, such as a migration to Hyperforce, while the dataset is being prepared for Analytics dashboards.

This error can occur if your org has been migrated to Hyperforce. If it has been more than 24 hours after the migration and your dataset hasn't self-replicated, reach out to Support to see if additional steps are needed. For more information, see Hyperforce - General Information & FAQ.

Field is not available
One or more fields are not visible to the Analytics Cloud Integration User, either because the field has been deleted or is not visible to the Analytics Cloud Integration User profile.

InternalServerError : Retried more than 15 times / InternalServerError: Retried more than 30 times
The Bulk API job created by Data Sync is too large. Removing text fields with large values from the Data Sync configuration will resolve the error.

Why is my recipe or data sync job running out of time or being killed?
A recipe or data sync job can fail because the integration user can't authenticate. If you receive one of these error messages, it typically indicates a temporary lack of server resources - rerunning the job usually resolves the issue.

NUMBER_OUTSIDE_VALID_RANGE
A modified Analytics Cloud Integration User profile does not contain "View All Data" or read access on a digested object.

Object is not available
The Analytics Cloud Integration User does not have access to an object, often because it is part of a managed package to which the Integration User does not have a license.

Queue wait time exceeded limit
Often a transient error that can be remedied by re-running the dataflow, or an authentication problem with the Integration User.

Query is too large. Must partition
The SOQL query generated from digesting an object contains too many characters.

Frequently Asked Questions 

What's the difference between a warning and an error?

A warning means the dataflow completed and all datasets registered, though a data quality issue may exist. An error means the dataflow failed entirely, and no datasets were registered.