Documentationbreadcrumb arrow Pluginsbreadcrumb arrow Databricks data sourcebreadcrumb arrow Annotations
Enterprise Grafana Cloud

Databricks annotations

Annotations in Grafana allow you to overlay event data on your graphs, making it easier to correlate metrics with specific events like deployments, incidents, or data pipeline runs. For an overview of annotations, refer to Annotate visualizations.

To create an annotation your dashboard must already be saved. Annotations only work with the following supported visualization types: time series, state timeline and candlestick

Examples:

Show data quality issues detected in your pipelines:

SQL 
SELECT 
  check_timestamp AS time,
  CONCAT('Data Quality Alert: ', table_name) AS title,
  CONCAT('Check: ', check_name, '\nExpected: ', expected_value,
    '\nActual: ', actual_value, '\nSeverity: ', severity) AS text,
  CONCAT('data-quality,', severity) AS tags
FROM monitoring.quality.check_results
WHERE check_timestamp >= $__from
  AND check_timestamp <= $__to
  AND status = 'FAILED'
ORDER BY check_timestamp

Track when Databricks jobs complete to correlate with performance metrics:

SQL 
SELECT 
  end_time * 1000 AS time,
  job_name AS title,
  CONCAT('Job ID: ', job_id, '\nStatus: ', state, '\nDuration: ', 
    ROUND((end_time - start_time) / 60, 2), ' minutes') AS text,
  CASE 
    WHEN state = 'SUCCESS' THEN 'green'
    WHEN state = 'FAILED' THEN 'red'
    ELSE 'orange'
  END AS tags
FROM system.workflow.job_runs
WHERE end_time >= UNIX_TIMESTAMP($__from) / 1000
  AND end_time <= UNIX_TIMESTAMP($__to) / 1000
ORDER BY end_time

Track cluster lifecycle events to understand resource usage:

SQL 
SELECT 
  event_time * 1000 AS time,
  pipeline_name AS title,
  CONCAT('Error: ', error_message, '\nPipeline ID: ', pipeline_id) AS text,
  'failure,alert' AS tags
FROM system.pipeline.events
WHERE event_type = 'FAILED'
  AND event_time >= UNIX_TIMESTAMP($__from) / 1000
  AND event_time <= UNIX_TIMESTAMP($__to) / 1000
ORDER BY event_time

Once configured, annotations appear as:

  • Vertical lines on your time series/candlestick/state timeline panels.
  • At the timestamp.
  • With hover tooltips showing the title and detailed text.
  • Filterable by the tags (like “data-quality” or severity level).