Neo4j Connector | OpenMetadata Database Integration

This section provides guides and references to use the Neo4j connector. Configure and schedule Neo4j metadata workflows from the OpenMetadata UI:

• Requirements
  • Entity Mapping
  • Data Type Mapping
• Metadata Ingestion
• Troubleshooting

​

Requirements

OpenMetadata reads the graph schema of one Neo4j database. It reads schema information only and never reads node or relationship data. The connecting user needs the ACCESS privilege on the database and permission to execute the following procedures:

• db.info
• db.labels
• db.relationshipTypes
• db.schema.nodeTypeProperties
• db.schema.relTypeProperties

The built-in reader role is enough.

​

Entity Mapping

OpenMetadata maps the Neo4j graph schema to OpenMetadata entities as follows: A property is NOT NULL only when every node type with that label has the property and marks it as mandatory. When a property has more than one type, or a type that OpenMetadata doesn’t recognize, OpenMetadata sets the column to the UNKNOWN data type and ingestion continues.

​

Data Type Mapping

OpenMetadata maps Neo4j property types to OpenMetadata data types as follows:

​

Metadata Ingestion

To ingest metadata from Neo4j, you need to create a service connection. The service connects Neo4j with OpenMetadata. After you create the service and deploy the metadata agent, OpenMetadata starts ingesting metadata.

​

Step 1: Add New Service

1. Navigate to Settings > Services.
2. Click Add New Service.

​

Step 2: Select a Service and Connector

From the service type dropdown, select Database Services, then click the Neo4j connector tile.

​

Step 3: Add Service Name and Description

• Enter a unique, descriptive Service Name. OpenMetadata identifies services by their service name. Enter a name that distinguishes this deployment from other Neo4j services you’re ingesting metadata from.
• Optional: Enter a Description for the service.

​

Step 4: Configure Connection Options

Specify your source credentials and verify the connection.

​

Enter Connection Details

Enter the connection details for Neo4j. The inline help panel in the UI describes each field. The form groups the fields into Connection, Authentication, Scope & Options, and Advanced Config sections.

• Host and Port (required): The host and port of the Neo4j server, for example localhost:7687. If you omit the port, OpenMetadata uses 7687.
• Username (required): The username to connect to Neo4j. The user needs read access to the graph schema of the database.
• Password (required): The password to connect to Neo4j.

Expand Scope & Options and Advanced Config to set the optional fields:

• Database Name (optional): The Neo4j database to extract metadata from. When left blank, OpenMetadata uses the home database of the user.
• Property Filter Pattern (optional): A regex that includes or excludes the node and relationship properties (columns) that match the pattern. Use it, for example, to drop properties that graph algorithms write back, such as louvain.*.
• Include Relationships: On by default. When on, OpenMetadata also ingests relationship types as tables in the relationships schema, with their properties as columns.

• Connection Scheme (optional): The Neo4j driver URI scheme, in Advanced Config. Defaults to neo4j. Choose one of neo4j, neo4j+s, neo4j+ssc, bolt, bolt+s, or bolt+ssc. A neo4j scheme makes a routed connection to a cluster or a single server, and a bolt scheme makes a direct connection to one server. A +s scheme encrypts the connection with full certificate validation. A +ssc scheme encrypts the connection and accepts self-signed certificates. Use neo4j+s for Neo4j Aura.

​

Test Connection

After you add the credentials, click Test Connection, then click Save. The test checks that OpenMetadata can connect to Neo4j, resolve the database, and read the node labels. It also reads the relationship types, but that step is optional.

​

Step 5: Configure Ingestion Options

In the What to Ingest step, configure filter rules for schemas and tables. The connector reads only the database set by Database Name, so database filter rules have no effect.

​

How Filter Rules Work

Each rule matches asset names using one of five match types:

• Contains: Matches any name containing the value. For example, sales matches my_sales_data and sales_2024.
• Starts with: Matches names beginning with the value. For example, prod_ matches prod_db and prod_schema.
• Ends with: Matches names ending with the value. For example, _raw matches events_raw and logs_raw.
• Is exactly: Matches the exact name only. For example, analytics matches only analytics.
• Matches regex: Matches names using a regular expression. For example, ^prod_.*_v\d+$ matches prod_events_v1.

When include rules and exclude rules both match an asset, the exclude rule takes priority.

​

Filter Options

The Database, Schema, and Table sections control ingestion for the corresponding asset type. Each section provides the following controls:

• Scan Mode: Choose one of the following scan modes:
  • Scan all: Ingests every asset of that type the connector can access. This is the default.
  • Only specific: Enables include rules so only assets matching at least one rule are ingested.
• Exclude system toggle: Use this toggle to automatically filter out system-reserved names defined by the connector, for example, Exclude system databases for the Databases section.
• Always exclude: Add permanent exclusion rules (shown in red). OpenMetadata never ingests assets matching these rules, regardless of include rules.
• Preview: View a real-time summary of what’s in scope based on your current rules.
• Include rules: In Only specific mode, click + Add to define a rule. Added rules appear as chips. OpenMetadata includes an asset if it matches any rule.

​

Step 6: Create & Deploy

Click Create & Deploy to deploy the agent and start the first metadata ingestion run. OpenMetadata saves the service configuration and immediately begins pulling metadata from the source. To monitor ingestion progress or view the service you just added, go to Settings > Services and select your service.

​

Configure Metadata Agent and Schedule Ingestion

The Metadata Agent extracts schemas, tables, columns, and other structural metadata from your source and keeps your OpenMetadata catalog in sync. It powers discovery, lineage, and governance across your data assets. When you click Create & Deploy, OpenMetadata automatically deploys a Metadata Agent for this service and triggers the first ingestion run. View its status and run history from the Agents tab on the service detail page. To configure the additional Metadata Agent and schedule ingestion, follow these steps:

1. Navigate to Settings > Services and select the service type.
2. Click the service you have added.
3. Select the Agents tab and click Add Agent > Metadata. For some services, the dropdown is not available and clicking Add Agent takes you directly to the agent configuration page.
4. On the Configure Ingestion page, do the following and click Next.
   • Name this Ingestion: Enter a unique recognizable name for this ingestion pipeline.
   • Agent Setup: Configure core parameters for metadata extraction. The following fields are available:

     Field	Default	Description
     Default Owner	—	Owner applied to any entity that has no owner from a more specific field below. Accepts a user or a team, by name or email.
     Service Owner	—	Owner assigned to the service entity itself (the connector/service you’re setting up, not the data inside it).
     Database Owner	—	Owner assigned to every database this agent ingests. Accepts either one owner for all of them, or a mapping of specific owners per database.
     Database Schema Owner	—	Owner assigned to every schema this agent ingests. Accepts either one owner for all of them, or a mapping of specific owners per schema.
     Table Owner	—	Owner assigned to every table this agent ingests. Accepts either one owner for all of them, or a mapping of specific owners per table.
     Enable Inheritance	On	When on, an entity with no owner of its own inherits one from the nearest parent that has one, checked in this order: table, then schema, then database, then service, then Default Owner. Turn off if you want only the owners set explicitly above to apply, with no fallback.
     Query Log Duration	1	How many days of query logs to look back through when processing stored procedure results.
     Query Parsing Timeout Limit	300	How many seconds to spend parsing a single query before giving up on it and moving to the next. Raise this if large or complex queries are being skipped.
     Number of Threads	1	How many tables to ingest in parallel. Raising this can speed up ingestion on databases with many tables, at the cost of more load on the source.
     Incremental Extraction	Off	When on, a run extracts only entities that changed since the last successful run, instead of re-scanning everything every time. Faster on large, mostly unchanged databases.
     Successful Pipeline Run Lookback Days	7	Only used when Incremental Extraction is on. How many days back to search for a prior successful run to use as the starting point for what changed since then.
     Safety Margin Days	1	Only used when Incremental Extraction is on. Extra days added before that starting point, as a buffer so changes still in flight when the last run finished aren’t missed.
     Extract JSON Schema	Off	Sample values in JSON columns to infer and ingest their schema, so the individual fields inside a JSON column are visible in OpenMetadata rather than just “JSON”. Requires SELECT permission on the sampled tables. If that permission is missing, the JSON column is still ingested, just without a detailed schema.
     JSON Schema Sample Size	10	Only used when Extract JSON Schema is on. Number of rows sampled per JSON column to infer its schema. A larger sample gives more accurate results but takes longer to run.
     Max Schema Inference Depth	No limit	Maximum nesting depth of the child columns OpenMetadata infers from sampled JSON values. The direct children of a column are depth 1. Deeper children are not ingested, and a column at the limit keeps its data type without children. 0 keeps no inferred children. Leave empty for no limit.
     Max Children Per Column	No limit	Maximum number of direct children kept for any inferred column, applied at every nesting level. When a column has more, OpenMetadata keeps the children whose names sort first, so every run keeps the same children. 0 keeps no inferred children. Leave empty for no limit.

     The limits apply to child columns OpenMetadata infers from sampled JSON values: JSON files and JSONL files in data lakes, JSON values inside CSV and TSV cells, NoSQL documents, and JSON columns when Extract JSON Schema is on. Declared schemas, such as JSON Schema, Avro, Parquet, Iceberg, and Delta metadata, are never limited. OpenMetadata logs one warning per table that names the affected columns. For more information about how the Default Owner, Service Owner, Database Owner, Database Schema Owner, Table Owner, and Enable Inheritance fields resolve an owner, see Hierarchical Owner Configuration.
   • Filter Patterns: Apply include or exclude rules to scope which databases, schemas, tables, and stored procedures this agent ingests. For more information about various filter options, see Step 6: Configure Ingestion Options.
   • Scope & Behaviour: Control how the agent handles metadata during ingestion. Toggle each option on or off based on your needs:

     Toggle	Default	Description
     Include Tables	On	Ingest tables and their columns. Turn off to skip tables entirely, for example if this agent exists only to pull tags or stored procedures.
     Include Tags	On	Ingest tags already defined on tables/columns in the source system (for example, a classification tag applied in Snowflake) and bring them into OpenMetadata as tags on the same assets.
     Include Stored Procedures	On	Ingest stored procedures as their own assets, so they show up in the catalog and can be linked into lineage.
     Include DDL Statements	Off	Also ingest the raw CREATE TABLE / CREATE VIEW statement for each asset, so it’s viewable in OpenMetadata alongside the asset’s metadata.
     Include Owners	Off	When an ingested asset’s owner in the source system is an email that matches an existing OpenMetadata user, assign that user as the asset’s owner. Never overwrites an owner an asset already has in OpenMetadata.
     Include Custom Properties	Off	Ingest source-specific metadata that doesn’t map to a standard OpenMetadata field (for example, a database-specific attribute) into custom properties on the asset, if custom properties have been defined for that entity type.
     Mark Deleted Tables	On	If a table this agent previously ingested no longer exists in the source, soft-delete it in OpenMetadata too. Only affects tables within schemas this agent actually scans.
     Mark Deleted Stored Procedures	On	Same as Mark Deleted Tables, but for stored procedures.
     Mark Deleted Schemas	Off	If an entire schema this agent previously ingested no longer exists in the source, soft-delete the schema and everything in it.
     Mark Deleted Databases	Off	If an entire database this agent previously ingested no longer exists in the source, soft-delete the database and everything in it.
     Override Metadata	Off	On: values from the source (descriptions, tags, owners, display names) overwrite what’s currently in OpenMetadata, even if someone edited it there. A description or display name that the source doesn’t send never replaces an existing value, and existing table constraints are kept when the source sends none. Override replaces tags only within the classifications the source sends, so tags from other classifications and glossary terms are kept. Retention periods that users set are kept. If the connector supplies custom properties, the values it sends are merged over the existing ones: properties the source sends overwrite, and properties it doesn’t send are kept. Off: OpenMetadata only fills in fields that are still empty, so manual edits in OpenMetadata are preserved.
     Enable Debug Log	Off	Run this ingestion at DEBUG log verbosity instead of the normal level. Turn on when troubleshooting a failing or unexpected run. Leave off otherwise, since debug logs are much larger.

     Available toggles vary by connector. Stored procedure options only appear for connectors that support stored procedures.
     
   • Advanced Config: Driver-level options (connection arguments, scheme, and timeouts). Most connections never need these, and they vary by connector. This section also has one ingestion-scope toggle:

     Toggle	Default	Description
     Include Views	On	Ingest views in addition to tables. Turn off to skip views entirely. Since view lineage is derived from parsing each view’s SQL definition, turning this off also stops view-based lineage from being generated for this service.

     
5. On the Schedule Interval page, set when the agent runs:
   • Schedule: Choose a preset interval (Hourly, Daily, Weekly, Monthly) or enter a custom cron expression.
   • On-Demand: No automatic schedule; trigger the agent manually when needed.
   
6. Click Add to deploy the agent.

​

Related