Neo4j Streams - Sink: Kafka → Neo4j

The Kafka Connect Neo4j Connector is the recommended method to integrate Kafka with Neo4j, as Neo4j Streams is no longer under active development and will not be supported after version 4.4 of Neo4j.

The most recent version of the Kafka Connect Neo4j Connector can be found here.

Is the Kafka Sink that ingest the data directly into Neo4j

How it works

It works in several ways:

  • by providing a Cypher template

  • by ingesting the events emitted from another Neo4j instance via the Change Data Capture module

  • by providing a pattern extraction to a JSON or AVRO file

  • by managing a CUD file format

The Cypher Template strategy is the only Sink strategy that guarantees messages to be processed in the same order as they arrive in a topic.

Other Sink strategies group messages together by type of operation, which can also be optimised into batches. In this case, the execution order is the following:

  1. All MERGE operations on nodes

  2. All DELETE operations on nodes

  3. All MERGE operations on relationships

  4. All DELETE operations on relationships

Cypher Template

It works with template Cypher queries stored into properties with the following format:

neo4j.conf
streams.sink.topic.cypher.<TOPIC_NAME>=<CYPHER_QUERY>

Each Cypher template must refer to an event object that will be injected by the Sink

Following an example:

For this event

{
 "id": 42,
 "properties": {
   "title": "Answer to anyting",
   "description": "It depends."
 }
}
neo4j.conf
streams.sink.topic.cypher.my-topic=MERGE (n:Label {id: event.id}) \
ON CREATE SET n += event.properties

Under the hood the Sink inject the event object as a parameter in this way

UNWIND {events} AS event
MERGE (n:Label {id: event.id})
    ON CREATE SET n += event.properties

Where {events} is a json list, so continuing with the example above a possible full representation could be:

:params events => [{id:"alice@example.com",properties:{name:"Alice",age:32}},
    {id:"bob@example.com",properties:{name:"Bob",age:42}}]

UNWIND {events} AS event
MERGE (n:Label {id: event.id})
    ON CREATE SET n += event.properties

When you decide to use Cypher template as Sink strategy to import data from Kafka into Neo4j, you have to be sure about the query correctness. If the query is not optimized, this could also results into possible performance issue or in situations where the plugin seems to be stuck, for example if the query loads a large amount of nodes and relationships into memory. We suggests the following:

  • execute a query EXPLAIN in order to better analyze the query and avoid this kind of situations

  • if Neo4j seems to be stuck then, from the Neo4j Browser, execute a CALL dbms.listQueries() to view all queries that are currently executing within the instance, and to be sure that there are no locked queries

Change Data Capture Event

This method allows to ingest CDC events coming from another Neo4j Instance. You can use two strategies:

  • The SourceId strategy which merges the nodes/relationships by the CDC event id field (it’s related to the Neo4j physical ID)

  • The Schema strategy which merges the nodes/relationships by the constraints (UNIQUENESS, NODE_KEY) defined in your graph model

The SourceId strategy

You can configure the topic in the following way:

streams.sink.topic.cdc.sourceId=<list of topics separated by semicolon>
streams.sink.topic.cdc.sourceId.labelName=<the label attached to the node, default=SourceEvent>
streams.sink.topic.cdc.sourceId.idName=<the id name given to the CDC id field, default=sourceId>
streams.sink.topic.cdc.sourceId=my-topic;my-other.topic

Each streams event will be projected into the related graph entity, for instance the following event:

{
  "meta": {
    "timestamp": 1532597182604,
    "username": "neo4j",
    "tx_id": 3,
    "tx_event_id": 0,
    "tx_events_count": 2,
    "operation": "created",
    "source": {
      "hostname": "neo4j.mycompany.com"
    }
  },
  "payload": {
    "id": "1004",
    "type": "node",
    "after": {
      "labels": ["Person"],
      "properties": {
        "email": "annek@noanswer.org",
        "last_name": "Kretchmar",
        "first_name": "Anne Marie"
      }
    }
  },
  "schema": {
    "properties": {
      "last_name": "String",
      "email": "String",
      "first_name": "String"
    },
    "constraints": [{
      "label": "Person",
      "properties": ["first_name", "last_name"],
      "type": "UNIQUE"
    }]
  }
}

will be persisted as the following node:

Person:SourceEvent{first_name: "Anne Marie", last_name: "Kretchmar", email: "annek@noanswer.org", sourceId: "1004"}

as you can notice, ingested event has been projected with two peculiarities:

  • the id field has transformed into sourceId;

  • the node has an additional label SourceEvent;

these two fields will be used in order to match the node/relationship for future updates/deletes

The Schema strategy

You can configure the topic in the following way:

streams.sink.topic.cdc.schema=<LIST_OF_TOPICS_SEPARATED_BY_SEMICOLON>
streams.sink.topic.cdc.schema=my-topic;my-other.topic

Each streams event will be projected into the related graph entity, for instance the following event:

{
  "meta": {
    "timestamp": 1532597182604,
    "username": "neo4j",
    "tx_id": 3,
    "tx_event_id": 0,
    "tx_events_count": 2,
    "operation": "created",
    "source": {
      "hostname": "neo4j.mycompany.com"
    }
  },
  "payload": {
    "id": "1004",
    "type": "node",
    "after": {
      "labels": ["Person"],
      "properties": {
        "email": "annek@noanswer.org",
        "last_name": "Kretchmar",
        "first_name": "Anne Marie"
      }
    }
  },
  "schema": {
    "properties": {
      "last_name": "String",
      "email": "String",
      "first_name": "String"
    },
    "constraints": [{
      "label": "Person",
      "properties": ["first_name", "last_name"],
      "type": "UNIQUE"
    }]
  }
}

will be persisted as the following node:

Person{first_name: "Anne Marie", last_name: "Kretchmar", email: "annek@noanswer.org"}

The Schema strategy leverages the schema field in order to insert/update the nodes so no extra fields will be created.

In case of relationship

{
  "meta": {
    "timestamp": 1532597182604,
    "username": "neo4j",
    "tx_id": 3,
    "tx_event_id": 0,
    "tx_events_count": 2,
    "operation": "created",
    "source": {
      "hostname": "neo4j.mycompany.com"
    }
  },
  "payload": {
    "id": "123",
    "type": "relationship",
    "label": "KNOWS",
    "start": {
      "labels": ["Person"],
      "id": "123",
      "ids": {
        "last_name": "Andrea",
        "first_name": "Santurbano"
      }
    },
    "end": {
      "labels": ["Person"],
      "id": "456",
      "ids": {
        "last_name": "Michael",
        "first_name": "Hunger"
      }
    },
    "after": {
      "properties": {
        "since": "2018-04-05T12:34:00[Europe/Berlin]"
      }
    }
  },
  "schema": {
    "properties": {
      "since": "ZonedDateTime"
    },
    "constraints": [{
      "label": "KNOWS",
      "properties": ["since"],
      "type": "RELATIONSHIP_PROPERTY_EXISTS"
    }]
  }
}

the Schema strategy leverages the ids fields in order to insert/update the relationships so no extra fields will be created.

The Pattern strategy

The Pattern strategy allows you to extract nodes and relationships from a json by providing a extraction pattern

Each property can be prefixed with:

  • !: identify the id (could be more than one property), it’s mandatory

  • -: exclude the property from the extraction If no prefix is specified this means that the property will be included

You cannot mix inclusion and exclusion so your pattern must contains all exclusion or inclusion properties

Labels can be chained via :

Tombstone Record Management

The pattern strategy come out with the support to the Tombstone Record, in order to leverage it your event should contain as key the record that you want to delete and null for the value.

Currently you can’t concatenate multiple patterns (for example in case you use just one topic and produce more then one node/relationship type). So you have to use a different topic for each type of node/relationship and define a pattern for each of them

The Node Pattern configuration

You can configure the node pattern extraction in the following way:

streams.sink.topic.pattern.node.<TOPIC_NAME>=<NODE_EXTRACTION_PATTERN>

So for instance, given the following json published via the user topic:

{"userId": 1, "name": "Andrea", "surname": "Santurbano", "address": {"city": "Venice", "cap": "30100"}}

You can transform it into a node by providing the following configuration:

by specifying a simpler pattern:

streams.sink.topic.pattern.node.user=User{!userId}

or by specifying a Cypher like node pattern:

streams.sink.topic.pattern.node.user=(:User{!userId})

Similar to the CDC pattern you can provide:

pattern meaning

User:Actor{!userId} or User:Actor{!userId,*}

the userId will be used as ID field and all properties of the json will be attached to the node with the provided labels (User and Actor) so the persisted node will be: (User:Actor{userId: 1, name: 'Andrea', surname: 'Santurbano', address.city: 'Venice', address.cap: 30100})

User{!userId, surname}

the userId will be used as ID field and only the surname property of the json will be attached to the node with the provided labels (User) so the persisted node will be: (User{userId: 1, surname: 'Santurbano'})

User{!userId, surname, address.city}

the userId will be used as ID field and only the surname and the address.city property of the json will be attached to the node with the provided labels (User) so the persisted node will be: (User{userId: 1, surname: 'Santurbano', address.city: 'Venice'})

User{!userId,-address}

the userId will be used as ID field and the address property will be excluded so the persisted node will be: (User{userId: 1, name: 'Andrea', surname: 'Santurbano'})

The Relationship Pattern configuration

You can configure the relationship pattern extraction in the following way:

streams.sink.topic.pattern.relationship.<TOPIC_NAME>=<RELATIONSHIP_EXTRACTION_PATTERN>

So for instance, given the following json published via the user topic:

{"userId": 1, "productId": 100, "price": 10, "currency": "€", "shippingAddress": {"city": "Venice", cap: "30100"}}

You can transform it into a path, like (n)-[r]→(m), by providing the following configuration:

By specifying a simpler pattern:

streams.sink.topic.pattern.relationship.user=User{!userId} BOUGHT{price, currency} Product{!productId}

or by specifying a Cypher like node pattern:

streams.sink.topic.pattern.relationship.user=(:User{!userId})-[:BOUGHT{price, currency}]->(:Product{!productId})

in this last case the we assume that User is the source node and Product the target node

Similar to the CDC pattern you can provide:

pattern meaning

(User{!userId})-[:BOUGHT]→(Product{!productId}) or (User{!userId})-[:BOUGHT{price, currency}]→(Product{!productId})

this will merge fetch/create the two nodes by the provided identifier and the BOUGHT relationship between them. And then set all the other json properties on them so the persisted data will be: (User{userId: 1})-[:BOUGHT{price: 10, currency: '€', shippingAddress.city: 'Venice', shippingAddress.cap: 30100}]→(Product{productId: 100})

(User{!userId})-[:BOUGHT{price}]→(Product{!productId})

this will merge fetch/create the two nodes by the provided identifier and the BOUGHT relationship between them. And then set all the specified json properties so the persisted pattern will be: (User{userId: 1})-[:BOUGHT{price: 10}]→(Product{productId: 100})

(User{!userId})-[:BOUGHT{-shippingAddress}]→(Product{!productId})

this will merge fetch/create the two nodes by the provided identifier and the BOUGHT relationship between them. And then set all the specified json properties (by the exclusion) so the persisted pattern will be: (User{userId: 1})-[:BOUGHT{price: 10, currency: '€'}]→(Product{productId: 100})

(User{!userId})-[:BOUGHT{price,currency, shippingAddress.city}]→(Product{!productId})

this will merge fetch/create the two nodes by the provided identifier and the BOUGHT relationship between them. And then set all the specified json properties so the persisted pattern will be: (User{userId: 1})-[:BOUGHT{price: 10, currency: '€', shippingAddress.city: 'Venice'}]→(Product{productId: 100})

Attach properties to node

By default no properties will be attached to the edge nodes but you can specify which property attach to each node. Given the following json published via the user topic:

{
    "userId": 1,
    "userName": "Andrea",
    "userSurname": "Santurbano",
    "productId": 100,
    "productName": "My Awesome Product!",
    "price": 10,
    "currency": "€"
}
pattern meaning

(User{!userId, userName, userSurname})-[:BOUGHT]→(Product{!productId, productName})

this will merge two nodes and the BOUGHT relationship between with all json properties them so the persisted pattern will be: (User{userId: 1, userName: 'Andrea', userSurname: 'Santurbano'})-[:BOUGHT{price: 10, currency: '€'}]→(Product{productId: 100, name: 'My Awesome Product!'})

CUD File Format

The CUD file format is JSON file that represents Graph Entities (Nodes/Relationships) and how to manage them in term of Create/Update/Delete operations.

You can configure the topic in the following way:

streams.sink.topic.cud=<LIST_OF_TOPICS_SEPARATED_BY_SEMICOLON>
streams.sink.topic.cud=my-topic;my-other.topic

We have two formats:

  • One for nodes:

    We provide an example of a MERGE operation

    {
      "op": "merge",
      "properties": {
        "foo": "value",
        "key": 1
      },
      "ids": {"key": 1, "otherKey":  "foo"},
      "labels": ["Foo","Bar"],
      "type": "node",
      "detach": true
    }

which would be transformed into the following Cypher query:

UNWIND [..., {
  "op": "merge",
  "properties": {
    "foo": "value",
    "key": 1
  },
  "ids": {"key": 1, "otherKey":  "foo"},
  "labels": ["Foo","Bar"],
  "type": "node",
  "detach": true
}, ...] AS event
MERGE (n:Foo:Bar {key: event.ids.key, otherkey: event.ids.otherkey})
SET n += event.properties

Lets describe the fields:

Table 1. CUD file Node format fields description
field mandatory Description

op

yes

The operation type: create/merge/update/delete

N.B. delete messages are for individual nodes it’s not intended to be a generic way of doing cypher query building from JSON

properties

no in case the operation is delete, otherwise yes

The properties attached to the node

ids

no in case the operation is create, otherwise yes

In case the operation is merge/update/delete this field is mandatory and contains the primary/unique keys of the node that will be use to do the lookup to the entity. In case you use as key the _id name the cud format will refer to Neo4j’s node internal for the node lookup.

N.B. If you’ll use the _id reference with the op merge it will work as simple update, this means that if the node with the passed internal id does not exists it will not be created.

labels

no

The labels attached to the node.

N.B. Neo4j allows to create nodes without labels, but from a performance perspective, it’s a bad idea don’t provide them.

type

yes

The entity type: node/relationship ⇒ node in this case

detach

no

In case the operation is delete you can specify if perform a "detach" delete that means delete any incident relationships when you delete a node

N.B. if no value is provided, the default is true

  • And one for relationships:

We provide an example of a CREATE operation

{
  "op": "create",
  "properties": {
    "foo": "rel-value",
    "key": 1
  },
  "rel_type": "MY_REL",
  "from": {
    "ids": {"key": 1},
    "labels": ["Foo","Bar"]
  },
  "to": {
    "ids": {"otherKey":1},
    "labels": ["FooBar"]
  },
  "type":"relationship"
}

which would be transformed into the following Cypher query:

UNWIND [..., {
  "op": "create",
  "properties": {
    "foo": "rel-value",
    "key": 1
  },
  "rel-type": "MY-REL",
  "from": {
    "ids": {"key": 1},
    "labels": ["Foo","Bar"]
  },
  "to": {
    "ids": {"otherKey":1},
    "labels": ["FooBar"]
  },
  "type":"relationship"
}, ...] AS event
MATCH (from:Foo:Bar {key: event.from.ids.key})
MATCH (to:FooBar {otherKey: event.to.ids.otherKey})
CREATE (from)-[r:MY_REL]->(to)
SET r = event.properties

Lets describe the fields:

Table 2. CUD file Relationship format fields description
field mandatory Description

op

yes

The operation type: create/merge/update/delete

properties

no

The properties attached to the relationship

rel_type

yes

The relationship type

from

yes, if you use the _id field reference into ids you can left labels blank

Contains the info about the source node of the relationship. For the description of the ids and labels fields please please look at the node fields description above

to

yes, if you use the _id field reference into ids you can left labels blank

Contains the info about the target node of the relationship. For the description of the ids and labels fields please please look at the node fields description above

type

yes

The entity type: node/relationship ⇒ relationship in this case

Following another example of DELETE operation for both node and relationship.

  • For Node, the following JSON:

{
  "op": "delete",
  "properties": {},
  "ids": {"key": 1, "otherKey":  "foo"},
  "labels": ["Foo","Bar"],
  "type": "node",
  "detach": false
}

will be transformed in the following Cypher query:

UNWIND [..., {
  "op": "delete",
  "properties": {},
  "ids": {"key": 1, "otherKey":  "foo"},
  "labels": ["Foo","Bar"],
  "type": "node",
  "detach": false
}, ...] AS event
MATCH (n:Foo:Bar {key: event.ids.key, otherkey: event.ids.otherkey})
DELETE n

Note that if you set "detach": true then the transformation will be:

UNWIND [
...
] AS event
...
DETACH DELETE n
  • For Relationship, the following JSON:

{
  "op": "create",
  "properties": {},
  "rel_type": "MY_REL",
  "from": {
    "ids": {"key": 1},
    "labels": ["Foo","Bar"]
  },
  "to": {
    "ids": {"otherKey":1},
    "labels": ["FooBar"]
  },
  "type":"relationship"
}

will be transformed in the following Cypher query:

UNWIND [..., {
  "op": "create",
  "properties": {},
  "rel_type": "MY_REL",
  "from": {
    "ids": {"key": 1},
    "labels": ["Foo","Bar"]
  },
  "to": {
    "ids": {"otherKey":1},
    "labels": ["FooBar"]
  },
  "type":"relationship"
}, ...] AS event
MATCH (from:Foo:Bar {key: event.from.ids.key})
MATCH (to:FooBar {otherkey: event.to.ids.otherkey})
MATCH (from)-[r:MY_REL]->(to)
DELETE r

We can create non-existent nodes at relationship creation/merging, putting "op": "merge" in "from" and/or "to" field. By default, "op" is match, so the node is not created if it doesn’t exist. We can write, for example:

{
  "op": "create",
  "properties": {},
  "rel_type": "MY_REL",
  "from": {
    "ids": {"key": 1},
    "labels": ["Foo","Bar"],
    "op": "merge"
  },
  "to": {
    "ids": {"otherKey":1},
    "labels": ["FooBar"],
    "op": "merge"
  },
  "type":"relationship"
}

How deal with bad data

The Neo4j Streams Plugin provides several means to handle processing errors.

It can fail fast or log errors with different detail levels. Another way is to re-route all the data and errors that for something reason it wasn’t able to ingest to a Dead Letter Queue.

It behaves by default like Kafka Connect, see this blog post
  • fail fast (abort) by default

  • need to configure dead-letter-queue topic to enable

  • need to enable logging explicitly

  • headers and message logging must be enabled explicitly

Config Options

Table 3. Dead Letter Queue configuration parameters
Name Value Note

errors.tolerance

none

fail fast (default!) abort

errors.tolerance

all

all == lenient, silently ignore bad messages

errors.log.enable

false/true

log errors (default: false)

errors.log.include.messages

false/true

log bad messages too (default: false)

errors.deadletterqueue.topic.name

topic-name

dead letter queue topic name, if left off no DLQ, default: not set

errors.deadletterqueue.context.headers.enable

false/true

enrich messages with metadata headers like exception, timestamp, org. topic, org.part, default:false

errors.deadletterqueue.context.headers.prefix

prefix-text

common prefix for header entries, e.g. "__streams.errors." , default: not set

errors.deadletterqueue.topic.replication.factor

3/1

replication factor, need to set to 1 for single partition, default:3

For the Neo4j extension you prefix them with streams.sink in the Neo4j configuration.

Example settings:

Fail Fast, Abort
errors.tolerance=none
Don’t fail on errors, Log with Messages
errors.tolerance=all
errors.log.enable=true
errors.log.include.messages=true
Don’t fail on errors, Don’t log but send to DLQ with headers
errors.tolerance=all
errors.deadletterqueue.topic.name=my-dlq-topic
errors.deadletterqueue.context.headers.enable=true
Same Settings for Neo4j Server Plugin
streams.sink.errors.tolerance=all
streams.sink.errors.deadletterqueue.topic.name=my-dlq-topic
streams.sink.errors.deadletterqueue.context.headers.enable=true

Every published record in the Dead Letter Queue contains the original record Key and Value pairs and optionally the following headers:

Header key Description

<prefix>topic

The topic where the data is published

<prefix>partition

The topic partition where the data is published

<prefix>soffset

The offset of the data into the topic partition

<prefix>class.name

The class that generated the error

<prefix>exception.class.name

The exception that generated the error

<prefix>exception.message

The exception message

<prefix>exception.stacktrace"

The exception stack trace

<prefix>databaseName"

The database name

Supported Kafka deserializers

The Neo4j Streams plugin supports 2 deserializers:

  • org.apache.kafka.common.serialization.ByteArrayDeserializer: if you want manage JSON messages

  • io.confluent.kafka.serializers.KafkaAvroDeserializer: if you want manage AVRO messages

You can define them independently for Key and Value as specified in the Configuration paragraph

Configuration summary

You can set the following Kafka configuration values in your neo4j.conf, here are the defaults.

neo4j.conf
kafka.bootstrap.servers=localhost:9092
kafka.auto.offset.reset=earliest
kafka.group.id=neo4j
kafka.enable.auto.commit=true
kafka.key.deserializer=org.apache.kafka.common.serialization.ByteArrayDeserializer
kafka.value.deserializer=org.apache.kafka.common.serialization.ByteArrayDeserializer

streams.sink.topic.cypher.<TOPIC_NAME>=<CYPHER_QUERY>
streams.sink.topic.cdc.sourceId=<LIST_OF_TOPICS_SEPARATED_BY_SEMICOLON>
streams.sink.topic.cdc.schema=<LIST_OF_TOPICS_SEPARATED_BY_SEMICOLON>
streams.sink.topic.cud=<LIST_OF_TOPICS_SEPARATED_BY_SEMICOLON>
streams.sink.topic.pattern.node.<TOPIC_NAME>=<NODE_EXTRACTION_PATTERN>
streams.sink.topic.pattern.relationship.<TOPIC_NAME>=<RELATIONSHIP_EXTRACTION_PATTERN>
streams.sink.enabled=<true/false, default=false>

streams.check.apoc.timeout=<ms to await for APOC being loaded, default -1 skip the wait>
streams.check.apoc.interval=<ms interval awaiting for APOC being loaded, default 1000>
streams.sink.poll.interval=<The delay interval between poll cycles, default 0>

See the Apache Kafka documentation for details on these settings.

if streams.cluster.only is set to true, streams will refuse to start in single instance mode, or when run in the context of the backup operation. This is an important safety guard to ensure that operations do not occur in unexpected situations for production deploys

See the Apache Kafka documentation for details on these settings.

Custom Kafka Configurations

In this section we describe the meaning of specific Neo4j streams Kafka configurations

kafka.streams.async.commit

If kafka.enable.auto.commit=false this property allows you to manage how to commit the messages to the topic.

Possible values:

  • false (default) under-the-hood we use the Kafka Consumer commitSync method

  • true under-the-hood we use the Kafka Consumer commitAsync method

commitSync VS commitAsync

commitSync is a synchronous commits and will block until either the commit succeeds or an unrecoverable error is encountered (in which case it is thrown to the caller).

That means, the commitSync is a blocking method with an interal retry mechanism, that can affect the performance of the ingestion because a new batch of messages will be processed only when the commit ended.

On the other hand commitAsync is an asynchronous call (so it will not block) and does not provide an internal retry mechanism.

Trade-offs: latency vs. data consistency

If you have to ensure the data consistency, choose commitSync because it will make sure that, before doing any further actions, you will know whether the offset commit is successful or failed. But because it is sync and blocking, you will spend more time on waiting for the commit to be finished, which leads to high latency. If you are ok of certain data inconsistency and want to have low latency, choose commitAsync because it will not wait to be finished.

Multi Database Support

Neo4j 4.0 Enterprise has multi-tenancy support, in order to support this feature you can set for each database instance a configuration suffix with the following pattern to.<DB_NAME> to the properties in your neo4j.conf file.

Following the list of new properties that allows to support multi-tenancy:

streams.sink.topic.cypher.<TOPIC_NAME>.to.<DB_NAME>=<CYPHER_QUERY>
streams.sink.topic.cdc.sourceId.to.<DB_NAME>=<LIST_OF_TOPICS_SEPARATE_BY_SEMICOLON>
streams.sink.topic.cdc.schema.to.<DB_NAME>=<LIST_OF_TOPICS_SEPARATE_BY_SEMICOLON>
streams.sink.topic.pattern.node.<TOPIC_NAME>.to.<DB_NAME>=<NODE_EXTRACTION_PATTERN>
streams.sink.topic.pattern.relationship.<TOPIC_NAME>.to.<DB_NAME>=<RELATIONSHIP_EXTRACTION_PATTERN>
streams.sink.enabled.to.<DB_NAME>=<true/false, default=false>

Please note the streams.sink.enabled.to.<DB_NAME> property. It is false by default because of the Sink module is disabled as well by default. In case the Sink module is enabled (streams.sink.enabled=true), then it will be enabled for every databases. So if you want to enable/disable the Sink module just for one or more databases you have to specify the streams.sink.enabled.to.<DB_NAME> property for each database.

This means that for each db instance you can specify if:

  • use the source connector

  • the routing patterns

So if you have a instance name foo you can specify a configuration in this way:

streams.sink.topic.cypher.<TOPIC_NAME>.to.foo=<CYPHER_QUERY>
streams.sink.topic.cdc.sourceId.to.foo=<LIST_OF_TOPICS_SEPARATE_BY_SEMICOLON>
streams.sink.topic.cdc.schema.to.foo=<LIST_OF_TOPICS_SEPARATE_BY_SEMICOLON>
streams.sink.topic.pattern.node.<TOPIC_NAME>.to.foo=<NODE_EXTRACTION_PATTERN>
streams.sink.topic.pattern.relationship.<TOPIC_NAME>.to.foo=<RELATIONSHIP_EXTRACTION_PATTERN>
streams.sink.enabled.to.foo=<true/false, default=false>

The old properties:

streams.sink.topic.cypher.<TOPIC_NAME>=<CYPHER_QUERY>
streams.sink.topic.cdc.sourceId=<LIST_OF_TOPICS_SEPARATE_BY_SEMICOLON>
streams.sink.topic.cdc.schema=<LIST_OF_TOPICS_SEPARATE_BY_SEMICOLON>
streams.sink.topic.pattern.node.<TOPIC_NAME>=<NODE_EXTRACTION_PATTERN>
streams.sink.topic.pattern.relationship.<TOPIC_NAME>=<RELATIONSHIP_EXTRACTION_PATTERN>
streams.sink.enabled=<true/false, default=false>

are still valid and they refer to Neo4j’s default db instance, which is usually called neo4j, but can be controlled by separate Neo4j system configuration.

The default database is controlled by Neo4j’s dbms.default_database configuration property so we’re being clear about which default database applies for this user. Database names are case-insensitive and normalized to lowercase, and must follow Neo4j database naming rules. (Reference: https://neo4j.com/docs/operations-manual/current/manage-databases/configuration/#manage-databases-administration)

In particular the following property will be used as default values for non-default db instances, in case of the specific configuration params is not provided:

streams.sink.enabled=<true/false, default=false>

This means that if you have Neo4j with 3 db instances:

  • neo4j (default)

  • foo

  • bar

and you want to enable the Sink plugin on all instance you can simply omit any configuration about enabling it, you just need to provide the routing configuration for each instance:

streams.sink.topic.cypher.fooTopic.to.foo=MERGE (f:Foo{fooId: event.fooId}) SET c += event.properties
streams.sink.topic.cypher.barTopic.to.bar=MERGE (b:Bar{barId: event.barId}) SET c += event.properties
streams.sink.topic.cypher.barTopic.to.neo4j=MERGE (c:MyLabel{myId: event.myId}) SET c += event.properties

Otherwise if you want to enable the Sink plugin only on customers and products instances you can do it in this way:

streams.sink.enabled=false
streams.sink.enabled.to.foo=true
streams.sink.enabled.to.bar=true
streams.sink.topic.cypher.fooTopic.to.foo=MERGE (f:Foo{fooId: event.fooId}) SET c += event.properties
streams.sink.topic.cypher.barTopic.to.bar=MERGE (b:Bar{barId: event.barId}) SET c += event.properties

So in general if you have:

streams.sink.enabled=true
streams.sink.enabled.to.foo=false

Then sink is enabled on all databases EXCEPT foo (local overrides global)