Grafana

Applies to VAST Cluster 5.5.

Grafana can chart the Catalog and the audit log through Trino: Grafana sends SQL to Trino, and Trino reads the tables with the VAST connector. Two community dashboards are published on grafana.com, one for each table. Both ran on VAST Cluster 5.5 with Trino 478, Grafana 12.0.2 and the Trino data source plugin 1.2.0: every panel returned data.

What you need

• Trino with the VAST connector, set up as on the Trino page, with a key that can read the Catalog or the audit log (access).
• Grafana with the Trino data source plugin (trino-datasource) pointing at that Trino. The plugin is third-party; follow its own documentation. The published dashboards need Grafana 11.3 or later and plugin 1.0.7 or later.

The published dashboards

Import a dashboard by its ID or JSON file in Grafana, and choose your Trino data source. Then set the dashboard variables at the top:

• Trino Catalog: the Trino catalog for VAST, vast if you followed the Trino page. The list comes from SHOW CATALOGS.
• Directory Prefix (Catalog) or Filename Prefix (audit log): the path to report on, such as /projects/. In the Catalog dashboard it becomes the search_path filter of every panel. Left empty, every panel reads every row your key can see, so set it.
• User Prefix and Action Prefix (audit log): filter on the start of login_name and of rpc_type.

The panels' SQL already uses the table names in Trino and Spark. Before you rely on a panel, note what it assumes:

• The audit dashboard needs full paths. Every panel filters on path. If your cluster isn't set to log full paths, path is empty and the dashboard shows nothing. Without user names, users show as uid and the UID.
• "Data written" counts WRITE and PUT_OBJECT only. Copies and multipart uploads aren't in it.
• Directory panels in the Catalog dashboard count every element under a directory, directories included, and a file with several hard links counts once per name.
• Refresh interval. The Catalog changes only when a new snapshot is indexed, so refreshing a Catalog dashboard more often than the Catalog's snapshot interval just repeats the same queries. The audit log has new rows as operations happen, but each query you run against it is audited too, as many QUERY_DATA records, when your platform team audits VAST DataBase operations.

Build your own panels

Any query on the Trino and Spark page works as a Grafana panel. Capacity by owner or extension makes a bar chart; operations per minute from the audit log make a time series, one series per protocol, for example:

-- not-run
SELECT date_trunc('minute', time) AS minute, protocol, count(*) AS operations
FROM vast."vast-audit-log-bucket/vast_audit_log_schema".vast_audit_log_table
WHERE time >= current_timestamp - INTERVAL '1' HOUR
GROUP BY 1, 2
ORDER BY 1;

See also

• Trino and Spark.
• Capacity reports and the audit log.