Warehouse Sync v2 Troubleshooting Guide

When setting up Warehouse Sync, it is possible to encounter problems with one or more components of your configuration:

• Initial data warehouse connectivity
• SQL query syntax
• Pipeline issues related to security or how mParticle parses your database
• Problems with mapping data from your warehouse to user profiles in mParticle

Following are several troubleshooting guides for each of these problem categories. If you are still encountering issues after following the appropriate steps below, contact mParticle support.

Data warehouse connectivity

Connectivity issues are often the result of an incomplete or incorrect first-time configuration.

Common symptoms

• Receiving an error code when using the API to create a connection to a warehouse

Before troubleshooting, verify the following:

1. Your mParticle account representative has enabled Warehouse Sync for the account you are using.
2. You can successfully connect to your data warehouse outside of mParticle, using the same username and password.
3. You followed the set-up steps specific to your data warehouse. Simple mistakes or typos made during this phase may prevent Warehouse Sync from working.
4. All of the relevant mP IP addresses are whitelisted.

Troubleshooting steps

• Validate all the data warehouse parameters in the POST {baseURI}/connections API call. Are they correct for the data warehouse instance you are trying to connect to?
• Compare your actual data location pod, organization, account, and workspace ID values with the values you are supplying in your API calls.
• From another application, connect to your warehouse using the username and password you created or specified. Ensure those credentials are permitted to access the pertinent datasets, tables, compute warehouses, and storage integrations.

SQL syntax

Errors or incompatibility in the SQL syntax of your data model will return errors and prevent the sync from succeeding.

Common symptoms

• The report API returns an error message indicating there is a syntax issue in the SQL statement provided in the data model.

Before troubleshooting, run the SQL query outside of mParticle. If it doesn’t run successfully or return the expected results, the issue is likely in your query, independent of Warehouse Sync.

Troubleshooting steps

• If you receive an error after running the SQL query, remove the part of the query highlighted in the error message.

• Verify your SQL syntax. While most data warehouses support common SQL syntax, it is possible to encounter exceptions in SQL extension for your warehouse. For example:

  • Snowflake doesn’t match case-sensitive, explicit identifiers to case-insensitive statements. For example, the statement SELECT current_timestamp AS \"tstamp\" FROM tableXYZ ... " in your SQL query will fail if iterator_field is tstamp in your data model.

• Workaround 1, Remove the explicit identifier " ":

  • SELECT current_timestamp AS tstamp FROM tableXYZ ... "
  • iterator_field": "tstamp"

• Workaround 2, force UPPER CASE:

  • SELECT current_timestamp AS \”TSTAMP\" FROM tableXYZ ... "
  • "iterator_field": "TSTAMP"

• If the error is related to the timestamp field in the query, ensure that:

  • You specified the correct column name and data type in the data model configuration.
  • You are not using dynamically generated timestamp values. Each data warehouse and environment may treat these values differently in terms of data type.

Pipeline issues

Pipeline issues are typically caused by security problems, the timestamp field provided in the data model, or other factors with the environment.

Common symptoms

• The report API returns some type of error message. For example:

  • Error assuming the AWS_ROLE. Please verify the role and externalId are configured correctly in your AWS policy.
  • Insufficient permission to extract records
  • Insufficient privileges to operate on integration ‘MP_US2_5000170_244_S3’
  • Validation Error: Missing required columns scanned_timestamp_ms in source query
  • SQL compilation error: error line 1 at position 36 invalid identifier ‘TIMESTAMP’
  • The dag’s data_interval_start is more than 7 days in the past. Found 14 days to back-fill in a ScheduleInterval.Hourly schedule.
  • Too many rows in the source query. Found 100000000 rows

• The report API returns "successful_records": 0. For example:

  {
    "pipeline_id": "string",
    "status": "idle",
    "connection_status": "healthy",
    "data_model_status": "valid",
    "latest_pipeline_run": {
      "id": 0,
      "pipeline_id": "string",
      "type": "scheduled",
      "status": "success",
      "errors": [
        {
          "message": "string"
        }
      ],
      "logical_date": "2023-10-25T18:11:57.321Z",
      "started_on": "2023-10-25T18:11:57.321Z",
      "ended_on": "2023-10-25T18:11:57.321Z",
      "range_start": "2023-10-25T18:11:57.321Z",
      "range_end": "2023-10-25T18:11:57.321Z",
      "successful_records": 0,
      "failed_records": 0
    }
  }

Before troubleshooting, verify the following:

1. You followed the configuration steps specific to your data warehouse. Any small mistake or typo will prevent Data Warehouse Sync from working.
2. You specified the correct datatype for the timestamps of the rows you are syncing.
3. The timestamps of the database rows you are syncing are not set in the future.
4. You are not exceeding the Warehouse Sync API limits

Troubleshooting steps

• If you are dynamically generating timestamp values, try using a literal value in the table or view you are querying.

Data import or mapping issues

Importing and mapping problems usually result from incorrect mapping between data rows in the warehouse and user profiles or attributes in mParticle.

Common symptoms

• mParticle created new profiles for users instead of updating existing profiles
• mParticle added new attributes to a profile instead of updating existing attributes

Before troubleshooting, verify the following:

1. The column names in your SQL query match the column names on your user profiles in mParticle.
2. The column names match the reserved mParticle user or device identity column names. For more information, see reserved mParticle user.

Troubleshooting steps

1. Correlate the row to the event batch according to your profile strategy.

2. Provide mParticle support or your account representative with the event batch JSON object from your mParticle Livestream, or the MPID and batch ID for the event, as well as a CSV of the source data

   • You can run the query manually against your data warehouse to simulate what mParticle extracted.
   • mParticle support can then confirm that the data lines up the expected behavior based on your data model.

Specific table schema changes in Google BigQuery

If a table schema changes and validation is still occurring, you may need to wait 24 hours for the cache in BigQuery to clear and reset before trying again.