Database Connections — User Guide

Audience: Dagen users who add and manage connections to databases, warehouses, lakes, streaming systems, and object storage for their workspace.


Overview

Open Database Connections at /db-connections. Here you create, edit, delete, test, and extract metadata for saved connections. That metadata feeds schema inventory, Data Modeling, agents, Data Ingestion (where compatible), and other features.

A full type-by-type overview—including Iceberg catalogs and ingestion overlap—is in Supported data sources, warehouses, and metadata.

Your workspace may restrict which connection types appear (admin policy).


Page layout

Area Purpose
Header Title Database Connections, short description, and Add New Connection (disabled while you are editing another connection in the dialog).
Page alert Success or error messages for save/delete and similar operations.
Existing Connections Grid of connection cards (responsive layout).
Available Databases (Across Connections) List of database names discovered across connections, each labeled with its connection name.
Add / Edit dialog Modal form for new connections or edits; title switches between Add New DB Connection and Edit DB Connection.

Empty state: If you have no connections yet, you’ll see a prompt to add your first one.

Loading / errors: A spinner appears while connections load; fetch errors show a warning at the top.


What you can connect (summary)

Category Examples
Relational / MPP PostgreSQL, MySQL, Oracle, Amazon Redshift, Teradata, Hive
Cloud warehouses Snowflake, Google BigQuery
Streaming / SaaS / storage Kafka, Salesforce, Amazon S3, Google Cloud Storage, Azure Blob Storage
Lake Apache Iceberg — catalog modes include REST, AWS Glue, Hive Metastore, Nessie; storage on S3, GCS, or Azure
Other Apache Ozone

Connection tests are type-aware (including Iceberg, Kafka, and common object-storage patterns).


Supported database types — required / typical fields

The connection type dropdown includes the native types below (not only PostgreSQL through Hive). Exact labels in the UI may vary slightly by release. Streaming and object-storage types usually show extra fields (TLS, SASL, region, bucket, prefix, etc.)—follow the live form.

Type Category Required / typical fields
PostgreSQL Relational Host, Port, Database name, Username, Password
MySQL Relational Host, Port, Database name, Username, Password
Oracle Relational Host, Port, SID or Service name (toggle), Username, Password
Amazon Redshift Relational / MPP Cluster endpoint (host), Port, Database name, Username, Password
Teradata Relational Host, Port, Database name, Username, Password
Apache Hive Relational / MPP Host, Port, Database / metastore-related fields, Username, Password
Snowflake Cloud warehouse Account identifier, Warehouse, Database, Schema, Username, Password
Google BigQuery Cloud warehouse Project ID, Dataset, Service account key (JSON)
Apache Kafka Streaming Bootstrap servers; security (SASL, SSL/TLS); client authentication as shown in the form
Salesforce SaaS Instance / OAuth and API credentials (per form)
Amazon S3 Object storage Region; credentials (e.g. access key/secret or IAM-based); bucket and path options (per form)
Google Cloud Storage Object storage Project; credentials; bucket/path (per form)
Azure Blob Storage Object storage Storage account; container; SAS or connection string / key (per form)
Apache Iceberg Lake Catalog type (e.g. REST, AWS Glue, Hive Metastore, Nessie); catalog connection details; warehouse path; storage backend (S3, GCS, Azure); credentials for catalog and storage
Apache Ozone Other Ozone endpoints / volume; authentication (per form)

If a type is missing from your dropdown, the workspace may use allowed-types restrictions—ask an admin. For ingestion-only systems, see also Supported data sources and Data Ingestion.


Adding or editing a connection

  1. Click Add New Connection (or Edit on a card).
  2. Enter a connection name and choose the type.
  3. Complete the form fields for that type (see table above).
  4. Submit to save.

Edit: Opens the same form with existing values. While a connection is open for edit, Add New Connection may be unavailable until you finish or cancel.


Connection cards

Each card shows:

  • Icon and type — visual hint for the system you connected.
  • Name — your label for this connection.
  • Summary fields — type-specific (e.g. host/port/database for Postgres; account/warehouse for Snowflake; project/dataset for BigQuery).
  • Added — when the connection was created.

Actions on each card

Action What it does
Extract Metadata Runs a metadata scan for this connection. Shows progress on the button; success or error may appear on the card. On success, Available Databases refreshes.
Edit Opens the form with current settings.
Delete Removes the connection (confirm if prompted).

From elsewhere in the workspace you can also use a saved connection for ingestion, pipelines, data model flows, and exploration—those entry points vary by feature; starting from Data Ingestion or Data Modeling often lets you pick a connection you defined here.


Secret managers and vaults

For credential fields you can often use Secret Manager mode and reference an external vault (for example HashiCorp Vault) instead of storing secrets directly in the form.

The Secret / Vault Providers section (when present) lists configured vault integrations. Use Test Connection on a provider to verify it is reachable. Setup may depend on backend configuration—follow in-app guidance.


Testing connections

  • Use Test Connection (or the type-specific test, e.g. for BigQuery) before relying on a connection for production jobs.
  • Failures are usually credentials, network path (firewall, VPN, allow lists), or wrong endpoint/catalog for Iceberg and cloud warehouses.

Metadata extraction

  • Extract Metadata updates Dagen’s view of schemas and objects for that connection and powers downstream features.
  • Agents and tools generally discover schemas and tables with lighter calls before deeper extraction, so exploration stays efficient.
  • Structured extraction is strongest for common warehouse and relational paths (e.g. Snowflake, Redshift, PostgreSQL, Oracle); behaviour aligns with your connection type.

More context: Supported data sources — metadata.


Available databases

The Available Databases (Across Connections) list shows database names discovered from your connections, with each row tied to the connection it came from. If nothing appears, run Extract Metadata or confirm the connection test passes.


Troubleshooting

Symptom Likely cause What to try
Connection type missing in the form Workspace allowed types policy Ask a workspace admin
Add New Connection disabled Edit dialog is open Finish or cancel the current edit
Test fails Network, credentials, or firewall Verify host, ports, secrets; allow Dagen to reach the endpoint
Metadata empty or partial Permissions or empty catalog Grant metadata/read access; confirm schemas exist
Extract Metadata error on card Timeout or engine error Retry; narrow scope; check logs if your role allows
Vault option inactive No vault provider configured Configure providers per admin docs / in-app alerts

Related