Sign in to the Azure portal and find your search service .

On the left, expand Search Management > Indexers and select an indexer.

Under Execution History , select the status. All statuses, including Success, have details about the execution.

If there's an error, hover over the error message. A pane appears on the right side of your screen displaying detailed information about the error.

Transient errors

For various reasons, such as transient network communication interruptions, timeouts from long-running processes, or specific document nuances, it's common to encounter transient errors or warnings during indexer runs. However, these errors are temporary and should be resolved in subsequent indexer runs.

To manage these errors effectively, we recommend putting your indexer on a schedule , for instance, to run every five minutes, where the next run commences five minutes after completing the first run, adhering to the maximum runtime limit on your service. Regularly scheduled runs help rectify transient errors or warnings.

If an error persists over multiple indexer runs, it's likely not a transient issue. In such cases, refer to the list below for potential solutions.

Error properties

Error: Could not map output field 'xyz' to search index due to deserialization problem while applying mapping function 'abc'

The output mapping might have failed because the output data is in the wrong format for the mapping function you're using. For example, applying Base64Encode mapping function on binary data would generate this error. To resolve the issue, either rerun indexer without specifying mapping function or ensure that the mapping function is compatible with the output field data type. See Output field mapping for details.

Error: Could not execute skill

The indexer wasn't able to run a skill in the skillset.

Error: Could not execute skill because the Web API request failed

The skill execution failed because the call to the Web API failed. Typically, this class of failure occurs when custom skills are used, in which case you need to debug your custom code to resolve the issue. If instead the failure is from a built-in skill, refer to the error message for help with fixing the issue.

While debugging this issue, be sure to pay attention to any skill input warnings for this skill. Your Web API endpoint might be failing because the indexer is passing it unexpected input.

Error: Could not execute skill because Web API skill response is invalid

The skill execution failed because the call to the Web API returned an invalid response. Typically, this class of failure occurs when custom skills are used, in which case you need to debug your custom code to resolve the issue. If instead the failure is from a built-in skill, file a support ticket to get assistance.

Error: Type of value has a mismatch with column type. Couldn't store in 'xyz' column. Expected type is 'abc'

If your data source has a field with a different data type than the field you're trying to map in your index, you might encounter this error. Check your data source field data types and make sure they're mapped correctly to your index data types .

Error: Skill did not execute within the time limit

There are two cases under which you might encounter this error message, each of which should be treated differently. Follow the instructions below depending on what skill returned this error for you.

Built-in Azure AI services skills

Many of the built-in cognitive skills, such as language detection, entity recognition, or OCR, are backed by an Azure AI services API endpoint. Sometimes there are transient issues with these endpoints and a request will time out. For transient issues, there's no remedy except to wait and try again. As a mitigation, consider setting your indexer to run on a schedule . Scheduled indexing picks up where it left off. Assuming transient issues are resolved, indexing and cognitive skill processing should be able to continue on the next scheduled run.

If you continue to see this error on the same document for a built-in cognitive skill, file a support ticket to get assistance, as this isn't expected.

Custom skills

If you encounter a timeout error with a custom skill, there are a couple of things you can try. First, review your custom skill and ensure that it's not getting stuck in an infinite loop and that it's returning a result consistently. Once you have confirmed that a result is returned, check the duration of execution. If you didn't explicitly set a timeout value on your custom skill definition, then the default timeout is 30 seconds. If 30 seconds isn't long enough for your skill to execute, you can specify a higher timeout value on your custom skill definition. Here's an example of a custom skill definition where the timeout is set to 90 seconds:

The maximum value that you can set for the timeout parameter is 230 seconds. If your custom skill is unable to execute consistently within 230 seconds, you might consider reducing the batchSize of your custom skill so that it has fewer documents to process within a single execution. If you have already set your batchSize to 1, you need to rewrite the skill to be able to execute in under 230 seconds, or otherwise split it into multiple custom skills so that the execution time for any single custom skill is a maximum of 230 seconds. Review the custom skill documentation for more information.

Error: Could not 'MergeOrUpload' | 'Delete' document to the search index

The document was read and processed, but the indexer couldn't add it to the search index. This can happen due to:

In all these cases, refer to Supported Data types and Data type map for indexers to make sure that you build the index schema correctly and have set up appropriate indexer field mappings . The error message includes details that can help track down the source of the mismatch.

Error: Integrated change tracking policy cannot be used because table has a composite primary key

This applies to SQL tables, and usually happens when the key is either defined as a composite key or, when the table has defined a unique clustered index (as in a SQL index, not an Azure Search index). The main reason is that the key attribute is modified to be a composite primary key in a unique clustered index . In that case, make sure that your SQL table doesn't have a unique clustered index, or that you map the key field to a field that is guaranteed not to have duplicate values.

Error: Could not process document within indexer max run time

This error occurs when the indexer is unable to finish processing a single document from the data source within the allowed execution time. Maximum running time is shorter when skillsets are used. When this error occurs, if you have maxFailedItems set to a value other than 0, the indexer bypasses the document on future runs so that indexing can progress. If you can't afford to skip any document, or if you're seeing this error consistently, consider breaking documents into smaller documents so that partial progress can be made within a single indexer execution.

Error: Could not project document

This error occurs when the indexer is attempting to project data into a knowledge store and there was a failure on the attempt. This failure could be consistent and fixable, or it could be a transient failure with the projection output sink that you might need to wait and retry in order to resolve. Here's a set of known failure states and possible resolutions.

Error: The cognitive service for skill '<skill-name>' has been throttled

Skill execution failed because the call to Azure AI services was throttled. Typically, this class of failure occurs when too many skills are executing in parallel. If you're using the Microsoft.Search.Documents client library to run the indexer, you can use the SearchIndexingBufferedSender to get automatic retry on failed steps. Otherwise, you can reset and rerun the indexer .

Error: Expected IndexAction metadata

An 'Expected IndexAction metadata' error means when the indexer attempted to read the document to identify what action should be taken, it didn't find any corresponding metadata on the document. Typically, this error occurs when the indexer has an annotation cache added or removed without resetting the indexer. To address this, you should reset and rerun the indexer .

Warning: Skill input was invalid

An input to the skill was missing, it has the wrong type, or otherwise, invalid. You might see the following information:

Could not execute skill

Skill executed but may have unexpected results

Cognitive skills have required inputs and optional inputs. For example, the Key phrase extraction skill has two required inputs text , languageCode , and no optional inputs. Custom skill inputs are all considered optional inputs.

If necessary inputs are missing or if the input isn't the right type, the skill gets skipped and generates a warning. Skipped skills don't generate outputs. If downstream skills consume the outputs of the skipped skill, they can generate other warnings.

If an optional input is missing, the skill still runs, but it might produce unexpected output due to the missing input.

In both cases, this warning is due to the shape of your data. For example, if you have a document containing information about people with the fields firstName , middleName , and lastName , you might have some documents that don't have an entry for middleName . If you pass middleName as an input to a skill in the pipeline, then it's expected that this skill input is missing some of the time. You need to evaluate your data and scenario to determine whether or not any action is required as a result of this warning.

If you want to provide a default value for a missing input, you can use the Conditional skill to generate a default value and then use the output of the Conditional skill as the skill input.

Warning: Skill input 'languageCode' has the following language codes 'X,Y,Z', at least one of which is invalid.

One or more of the values passed into the optional languageCode input of a downstream skill isn't supported. This can occur if you're passing the output of the LanguageDetectionSkill to subsequent skills, and the output consists of more languages than are supported in those downstream skills.

Note that you can also get a warning similar to this one if an invalid countryHint input gets passed to the LanguageDetectionSkill. If that happens, validate that the field you're using from your data source for that input contains valid ISO 3166-1 alpha-2 two letter country codes. If some are valid and some are invalid, continue with the following guidance but replace languageCode with countryHint and defaultLanguageCode with defaultCountryHint to match your use case.

If you know that your data set is all in one language, you should remove the LanguageDetectionSkill and the languageCode skill input and use the defaultLanguageCode skill parameter for that skill instead, assuming the language is supported for that skill.

If you know that your data set contains multiple languages and thus you need the LanguageDetectionSkill and languageCode input, consider adding a ConditionalSkill to filter out the text with languages that aren't supported before passing in the text to the downstream skill. Here's an example of what this might look like for the EntityRecognitionSkill:

Here are some references for the currently supported languages for each of the skills that can produce this error message: