VW_DATASET_TEST_EXECUTIONS

Overview

VW_DATASET_TEST_EXECUTIONS is the most detailed view for analyzing test execution history. Each row represents one test execution and includes the result, timing, row-level counts, error messages, metadata links, and windowed index columns that make it easy to retrieve the latest or earliest execution for any test.

This view is the primary bridge between test definitions (REPOSITORY views) and detailed result data (RESULTS tables). Use it for trend analysis, SLA monitoring, and drilling into specific failures.

Schema

Execution Identity

Result

Timing

Row Counts

Test Context

Data Source Information

Metadata Links

Job/Batch Context

Quality Classification

Windowed Index Columns

These index columns are pre-calculated ROW_NUMBER() partitions ordered by DATE_STARTED. They eliminate the need for window functions in your queries.

Relationships

VW_DATASET_TEST_EXECUTIONS
    │
    ├── TEST_EXECUTION_ID → RESULTS.TEST_EXECUTION_RESULTS.TEST_EXECUTION_ID
    ├── TEST_EXECUTION_ID → RESULTS.TEST_EXECUTION_TABLE_RESULTS.TEST_EXECUTION_ID
    ├── TEST_EXECUTION_ID → RESULTS.TEST_EXECUTION_TABLE_RESULT_KEYS.TEST_EXECUTION_ID
    ├── TEST_EXECUTION_ID → RESULTS.TEST_EXECUTION_TABLE_RESULT_VALUES.TEST_EXECUTION_ID
    ├── TEST_ID → VW_DATASET_TESTS.TEST_ID
    ├── PROJECT_ID → VW_DATASET_JOBS.PROJECT_ID
    ├── BATCH_ID → VW_DATASET_JOB_EXECUTIONS.BATCH_ID
    └── TEST_PRIMARY_METADATA_LINK_OBJECT_KEY → VW_DATASET_TABLES.TABLE_KEY

Example Queries

Latest Execution Per Test

SELECT
    TEST_NAME,
    RESULT,
    DATE_COMPLETED,
    DURATION,
    PASSED_COUNT,
    FAILED_COUNT,
    ERROR_COUNT,
    TRIGGERED_BY
FROM REPOSITORY.VW_DATASET_TEST_EXECUTIONS
WHERE LATEST_TEST_EXECUTION_INDEX = 1
ORDER BY DATE_COMPLETED DESC;

Test Execution Duration Trends

SELECT
    TEST_NAME,
    DATE_STARTED::DATE AS EXECUTION_DATE,
    DURATION,
    RESULT
FROM REPOSITORY.VW_DATASET_TEST_EXECUTIONS
WHERE TEST_NAME = 'Row Count - DIM_CUSTOMER'
    AND DATE_STARTED >= DATEADD(day, -30, CURRENT_DATE)
ORDER BY DATE_STARTED;

Failed Executions with Error Details

SELECT
    TEST_NAME,
    PROJECT_NAME,
    DATE_COMPLETED,
    RESULT,
    TEST_DATASET_ERROR,
    CONTROL_DATASET_ERROR,
    BATCH_ERROR,
    URL
FROM REPOSITORY.VW_DATASET_TEST_EXECUTIONS
WHERE RESULT = 'Error'
    AND DATE_COMPLETED >= DATEADD(day, -7, CURRENT_DATE)
ORDER BY DATE_COMPLETED DESC;

Join to Results for Row-Level Drill-Down

SELECT
    te.TEST_NAME,
    te.RESULT,
    r.NUMERIC_SOURCE_VALUE,
    r.NUMERIC_TARGET_VALUE,
    r.PERCENT_DIFFERENCE,
    r.EVALUATION_RESULT_NAME
FROM REPOSITORY.VW_DATASET_TEST_EXECUTIONS te
JOIN RESULTS.TEST_EXECUTION_RESULTS r
    ON r.TEST_EXECUTION_ID = te.TEST_EXECUTION_ID
WHERE te.LATEST_TEST_EXECUTION_INDEX = 1
    AND te.RESULT = 'Failed'
LIMIT 100;

Execution Summary by Job

SELECT
    RUNTIME_PARENT,
    DATE_STARTED::DATE AS RUN_DATE,
    COUNT(*) AS TOTAL_TESTS,
    SUM(CASE WHEN RESULT = 'Passed' THEN 1 ELSE 0 END) AS PASSED,
    SUM(CASE WHEN RESULT = 'Failed' THEN 1 ELSE 0 END) AS FAILED,
    SUM(CASE WHEN RESULT = 'Error' THEN 1 ELSE 0 END) AS ERRORED,
    AVG(DURATION) AS AVG_DURATION_SEC
FROM REPOSITORY.VW_DATASET_TEST_EXECUTIONS
WHERE DATE_STARTED >= DATEADD(day, -7, CURRENT_DATE)
GROUP BY RUNTIME_PARENT, DATE_STARTED::DATE
ORDER BY RUN_DATE DESC, RUNTIME_PARENT;

Databricks Notes

• Replace DATEADD(day, -30, CURRENT_DATE) with DATE_ADD(CURRENT_DATE(), -30)
• Replace DATE_STARTED::DATE with CAST(DATE_STARTED AS DATE) or DATE(DATE_STARTED)
• TIMESTAMPDIFF is not available — use UNIX_TIMESTAMP(end) - UNIX_TIMESTAMP(start) for duration if recalculating