Trino

Connecting Trino to Upriver

Upriver connects to your Trino cluster to monitor tables, run data quality checks, and support agent workflows over your lakehouse data. Access is read-only: Upriver queries table metadata and runs SELECT statements against the tables you configure—it does not modify data or schema objects.

To connect Trino, you will:

1. Create a dedicated Trino user with the minimum read permissions.

2. Ensure the Trino coordinator is reachable from Upriver over the network.

3. Register the cluster as a Trino integration in Upriver.

4. Create a Trino data source that points at the catalog, schema, and table to monitor.

---

Prerequisites

Network access

The Trino coordinator must be reachable from your Upriver deployment using the host and port you provide during setup (default coordinator port is 8080 unless your cluster uses another value).

• SaaS: Allow inbound connections from Upriver’s network to the coordinator (firewall, security group, or private link—coordinate with your Upriver representative if you need egress/ingress details).

• Self-hosted: Ensure routing and DNS resolve from the Upriver runtime to the coordinator hostname.

---

Step 1: Create a dedicated Trino user

Create a service account for Upriver (for example, upriver) rather than reusing a personal login. Grant read-only access only to the catalogs and schemas you intend to monitor.

Required capabilities

The Upriver user must be able to:

1. Query monitored tables — run SELECT (and limited metadata queries) on the catalog, schema, and table you configure in Upriver.

2. Read information_schema — Upriver introspects information_schema.tables and information_schema.columns in your catalog to discover schema and column metadata.

3. Read Iceberg partition metadata (Iceberg catalogs only) — for Iceberg tables, the user also needs access to the $partitions system table on monitored tables so Upriver can discover partition columns for time-based filtering.

Example grants

Exact syntax depends on your Trino version and access-control plugin (file-based, Ranger, OPA, etc.). Adapt catalog and schema names to your environment:

-- Example: create a dedicated user (syntax may vary by auth backend)
CREATE USER upriver WITH PASSWORD '<strong-password>';

-- Read access to the schema that contains monitored tables
GRANT SELECT ON SCHEMA iceberg.analytics TO upriver;

-- If your deployment uses table-level grants instead:
-- GRANT SELECT ON TABLE iceberg.analytics.orders TO upriver;

Work with your Trino administrator to ensure:

• The user cannot INSERT, UPDATE, DELETE, or CREATE objects outside what you explicitly allow.

• information_schema in the target catalog is readable for the schemas you monitor.

• For Iceberg, SELECT on "<catalog>"."<schema>"."<table>$partitions" is allowed when partition discovery is enabled.

---

Step 2: Add the Trino integration in Upriver

1. In Upriver, go to Settings → Integrations.

2. Click Add integration and choose Trino.

3. On the Connect step, review the user and network requirements above.

4. On the Finalize step, enter:

Field	Description
Host	Trino coordinator hostname or IP (required).
Port	Coordinator port (required; commonly 8080).
Username	The dedicated Trino user (required).
Password	Password for the user, if your cluster uses password authentication (optional when auth is disabled).

1. Save the integration.

Credentials are stored securely and keyed by coordinator host so you can reuse the same integration across multiple Trino data sources on that cluster.

---

Step 3: Create a Trino data source

After the integration exists, create a data source that identifies the table Upriver should monitor:

1. Go to Data sources and create a new source of type Trino.

2. Select the Trino integration you created.

3. Provide:

Field	Description
Catalog	Trino catalog name (for example, iceberg, hive, or delta).
Catalog type	iceberg, hive, or delta—must match how the catalog is implemented.
Schema	Schema containing the table.
Table	Table to monitor.

1. Complete the remaining data source wizard steps (scheduling, columns, etc.) as needed.

Upriver connects using the integration’s host, port, and username, and runs read queries of the form:

SELECT * FROM "<catalog>"."<schema>"."<table>" ...

---

Catalog-specific notes

Catalog type	Notes
Iceberg	Ensure read access to information_schema and to the $partitions metadata table for monitored Iceberg tables when partition-based time filtering is used.
Hive	Grant SELECT on the target schema or tables; partition behavior follows Hive layout and Upriver’s partition configuration for the data source.
Delta	Grant SELECT on the target schema or tables in the Delta catalog connector you use with Trino.