Trino

Integrating Trino and Starburst with Coalesce Quality

This guide will show you how to securely connect Coalesce Quality to your Trino cluster or Starburst Galaxy. We need this information so we collect metadata about your tables.

To be able to finish this guide, you’ll need the following:
→ Access to create a user for Coalesce Quality and to change what it may read in your Trino access control

⏱️ Estimated time to finish: 10 minutes.

​

Data we collect

We read metadata from the catalogs you list in the integration, each on its own, using the dedicated user described below:

• Catalog: schemas, tables, views, materialized views, columns and data types, from each catalog’s information_schema and system.metadata. Table comments only with Fetch Trino table comments on.
• SQL definitions: the CREATE statement of each table and view, from SHOW CREATE TABLE and SHOW CREATE VIEW, and the query of each materialized view. Used for lineage.
• Constraints: the keys and checks a connector reports in information_schema.
• Table metrics: the row count of each table, from SHOW STATS. For an Iceberg table (the iceberg connector, and Starburst Galaxy object storage), also the time of its last snapshot, from the table’s $snapshots metadata table.

Coalesce Quality does not collect Trino query history.

​

Setup Trino access

​

Create a user

Create a dedicated user for Coalesce Quality in the authentication your cluster uses (password file, LDAP, or a Starburst Galaxy service account) and note its username and password.

​

Grant read access

How you grant access depends on the access control of your cluster. The user needs, in each catalog you list:

• To see the schemas, tables and views, so they show in information_schema and system.metadata.
• SELECT on the tables. SHOW STATS and the Iceberg $snapshots table are read through it. Coalesce Quality reads table statistics and snapshot metadata, never the rows themselves, unless you run monitors that query your data.
• SHOW CREATE TABLE and SHOW CREATE VIEW on the tables and views, unless you turn those off below.

A table or view whose SHOW CREATE is refused is read without it, and everything else works. If your access control allows SHOW CREATE TABLE only to a table’s owner, turn on Do not use SHOW CREATE TABLE to get table DDLs below rather than give the user ownership of your tables.

​

Network access

For the complete list of IP addresses by region, see Coalesce Quality IP Whitelist.

Allow inbound HTTPS connections from the Coalesce Quality IPs for your region to the Trino coordinator.

​

Input data in the Coalesce Quality UI

​

Integration name

For example Trino.

​

Host and port

The host of your Trino coordinator, for example trino.example.com, and its port. Leave the port empty to use the default of the protocol.

​

Use non-SSL connection

Connects over plain HTTP instead of HTTPS. Use it only for a connection that never leaves a private network.

​

Username and password

The user you created above.

​

Catalogs

The catalogs to read. Only the catalogs listed here are read, so list every catalog you want to monitor. You can add filters to a catalog to restrict which schemas and tables are collected from it. A catalog that cannot be read, for example because its connector cannot reach the database behind it, is reported on the integration, and the other catalogs are still read.

​

Metrics scope (optional)

Restricts which tables are scanned for metrics. Leave it empty to collect metrics for all ingested tables.

​

Audit table FQN (optional)

Fully qualified name (catalog.schema.table) of the table where Coalesce Quality writes SQL test audit logs. Coalesce Quality creates the table if it does not exist. Leave empty if you don’t run SQL tests.

​

Data collection options

• Fetch table metrics (required for automatic monitors): on by default. Every 30 minutes, reads the row count of each table, and the last snapshot time of Iceberg tables. Automatic freshness and volume monitors stop working when it is off.
• Do not use SHOW CREATE VIEW to get view DDLs: reads the query of each view from information_schema.views instead.
• Do not use SHOW CREATE TABLE to get table DDLs: tables then have no DDL.
• Do not fetch Trino materialized views: skips system.metadata.materialized_views, for connectors that fail on it.
• Fetch Trino table comments: reads table comments from system.metadata.table_comments.

​

On-premise agent

On the on-premise agent, the same settings are keys of a trino connection:

connections:
  "trino-prod":
    trino:
      host: trino.example.com
      port: 443
      username: coalesce_quality
      password: ${TRINO_PASSWORD}
      catalogs:
        - iceberg
        - hive
      fetch_table_comments: true
      no_show_create_table: false
      no_show_create_view: false
      no_materialized_views: false

use_plaintext: true is the agent’s Use non-SSL connection.