Entity Resolution Recipe Troubleshooting

This page documents the known problems while running the Entity Resolution recipe and the steps to resolve them. Each section describes a symptom, its cause, and the resolution.

Selecting and Configuring

No Tables Are Listed on Select Asset

After picking a datastore and clicking Existing Table, the list is empty and reads "No profiled tables in this datastore. Run a profile operation first."

Cause: The recipe only offers tables that have been profiled. It reads the profile for the field list, the field metrics, and the AI suggestions, so an unprofiled table cannot be used.

Resolution: Run a Profile operation on the datastore, then reopen the step. A profile that has never run on a new table also explains why a table you can see elsewhere is missing here.

The Computed Table and Computed Join Cards Are Missing

Select Asset offers only Existing Table.

Cause: The two computed cards appear only when you can create computed assets on the datastore, which needs the Author team permission (as the asset's owner) or Editor.

Resolution: Use an existing profiled table, ask an Editor to create the computed asset for you and profile it, or request the team permission. See Permissions.

An AI Suggestion Could Not Load

The footer shows "Could not load the AI distinction field suggestion. You can still pick a distinction field manually." or the equivalent message for blocking fields, compare fields, the dry-run analysis, or the scan interpretation.

Cause: The AI request failed. The usual reasons are that the AgentQ data sharing level is below Metadata Shared, that the AI provider did not answer in time or returned a response Qualytics could not use, or that the table has no profiled fields.

Resolution: Click Retry. If it fails again, continue manually; every step works without AI. Ask your administrator to check the AgentQ data sharing level and the AI integration under Settings > Integrations. See AgentQ Access Controls.

No Field Suggestion Appeared

With AgentQ configured, no suggestion is shown on Blocking Fields or Compare Fields, and no error is shown either.

Cause: AgentQ reviewed the container and had nothing to propose for that role. A table of audit or lifecycle columns often has no pair of fields worth comparing and nothing useful to block on. Having nothing to suggest is a legitimate answer rather than a failure, so the step stays quiet instead of reporting an error, and retrying cannot change it.

Resolution: Configure that role yourself; every step works without AI. The two roles are asked for separately, so a suggestion can still arrive on one when the other had nothing to offer. Distinction Field behaves differently: it answers with a field or with the error in An AI Suggestion Could Not Load, never with silence.

Review Reports a Conflict with an Existing Check

The Review step shows Conflict with an existing check: "An active entity-resolution check already exists for this asset. Load it to continue, or change the asset above so it no longer conflicts." Next is unavailable.

Cause: An active check conflicts with the one you are building when it has the same rule type, the same asset, the same fields, and the same filter. The recipe found an active Entity Resolution check on the selected table that overlaps with your configuration this way.

Resolution: Click Load existing check to continue from the existing check's configuration, which discards the settings you were building. Or change the asset on Select Asset. To keep a new configuration on the same table, archive the existing check from the datastore page first, or change the compare fields or the filter so the two checks no longer overlap.

Validating and Scanning

The Dry Run Finds No Duplicates

Validate reports "No duplicates found in the sample. If you expected matches, lower the threshold or adjust the compare fields, then validate again."

Cause: The sampled dry run analyzes up to 10,000 records per partition. Either the sample contains no duplicates that the current configuration matches, or the threshold is too strict for the compare fields you chose. When AgentQ is available, the assessment No matches found; the threshold may be too strict points at the second case.

Resolution: Lower the threshold on Match Threshold, or switch a strict exact comparison to a fuzzy one on Compare Fields, and validate again. Iterating on the sample is quick. If you know the table holds duplicates that never appear in a sample, validate as is and rely on the full scan.

The Scan Step Asks to Link an Enrichment Destination

Scan shows "Choose a datastore to receive the scan's records" and a Link Enrichment button instead of Scan.

Cause: The source datastore has no linked enrichment destination. The scan stores the source records of the duplicates it finds there, and the golden set is written there too.

Resolution: Click Link Enrichment and complete the dialog. The step unblocks as soon as the link is saved. See Link an Enrichment Destination from the Scan Step.

The Scan or the Materialize Fails Right Away

The step shows an error with a Retry Scan or Retry action as soon as you click Scan or Materialize.

Cause: Either you lack the Editor team permission on the source datastore, which running operations requires, or, for the materialize, the enrichment destination linked to the source datastore is disconnected or its link was removed after the scan.

Resolution: Confirm your team permission on the datastore. Test the enrichment datastore's connection from its page and fix the credentials or network access, then click Retry. See Permissions.

Source Records Are Not Loaded

The Scan step shows Source records not yet loaded, or a later step reads "Source records are not loaded. Please go back to the Scan step and run a full scan to reload the data needed for the golden set."

Cause: The scan finished but the anomaly's source records were not retrieved, for example because the anomaly was still being written when the recipe looked for it, or because you resumed the recipe in a session where the cached records were gone.

Resolution: On the Scan step, click Load Source Records. If the button does not appear, click Run scan again. The records are needed to group the duplicates for remediation.

Remediating and Materializing

No AI Suggestion Arrived for an Entity

An entity card shows "No AI suggestion arrived for this entity." or the footer reads "Could not load AI recommendations for N entities. You can still select their golden records manually."

Cause: Golden-record recommendations are scored a few entities at a time, and the AI provider did not return a usable answer for some of them within the time allowed. Recommendations for the other entities are unaffected.

Resolution: Click Retry on the entity, or pick the record to keep yourself in the Keep as golden column. The recipe only requires at least one record per entity; it does not require an AI recommendation.

AI Suggestions Are Not Offered on Remediate

The AI suggestions are available banner never appears, or Get AI suggestions fails with "Could not load AI recommendations. You can still select golden records manually for each entity."

Cause: Golden-record recommendations send the records themselves to the model, so they need the AgentQ data sharing level Source Data Shared. Below that level the request is refused; the other AI steps, which send only metadata, keep working. When no AI provider is configured at all, no AI banner appears anywhere in the recipe.

Resolution: Select the golden records manually. To enable the recommendations, ask your administrator to raise the data sharing level; see AgentQ Access Controls.

The Selection Is Above the Limit of 25,000 Excluded Records

Materialize shows "This selection excludes N records, above the limit of 25,000. Narrow the scan or resolve these duplicates in batches." and Materialize is disabled.

Cause: The excluded records are passed to the materialize operation as a list of distinction-field values, and one run accepts at most 25,000 of them. A related message, naming the distinction field and saying its excluded values "are too large to send in one request. Resolve these duplicates in batches.", appears when the values themselves are too long for a single request.

Resolution: Resolve the duplicates in batches. Add a filter to the check so each scan covers part of the table, or keep more records per entity, then materialize and rerun through Scan in Recipe for the rest.

The Anomaly Was Not Acknowledged

The golden set was written but the message Anomaly not acknowledged appeared, and the anomaly is still Active.

Cause: Acknowledging the anomaly after materializing needs the Author team permission on the datastore, or the anomaly had been archived in the meantime.

Resolution: Acknowledge the anomaly from the Anomalies page once you have the permission. See Acknowledge Anomalies. The golden set is unaffected.

Resuming

The Recipe Restarted from the Beginning

Reopening Recipes shows Get Started instead of Continue, or the recipe opens on Select Asset although you had progressed further.

Cause: Progress is kept in your browser session, per recipe. A different browser or computer, a cleared session, or entering the recipe through Open in Recipe, Scan in Recipe, or Remediate in Recipe starts fresh.

Resolution: Nothing the run produced is lost. Open the active check and click Scan in Recipe to continue at the Scan step, or open its anomaly and click Remediate in Recipe to continue at Remediate. See Entry Points and Resuming.

Limitations

• The recipe works on one asset at a time. To resolve duplicates across two tables, build a Computed Join first.
• A run can exclude at most 25,000 records from the golden set. Larger duplicate sets are resolved in batches.
• The dry run analyzes up to 10,000 records per partition, so its counts are estimates of what the full scan will find.
• Progress is not shared between browsers or users. Everything the run produced is.