Neo4j 4.4

Human Connection is a social network. Using a graph based database which can model nodes and edges natively - a network - feels like an obvious choice. We decided to use Neo4j, the currently most used graph database available. The community edition of Neo4J is Free and Open Source and we try our best to keep our application compatible with the community edition only.

Installation With Docker

Run:

docker-compose up

You can access Neo4J through http://localhost:7474/ for an interactive Cypher shell and a visualization of the graph.

Installation Without Docker

Install the community edition of Neo4j along with the plugin Apoc on your system.

To do so, go to releases, choose "Community Server", download the installation files for you operation system and unpack the files.

Download Neo4j Apoc and drop the .jar file into the plugins folder of the just extracted Neo4j-Server.

Then make sure to allow Apoc procedures by adding the following line to your Neo4j configuration (conf/neo4j.conf):

dbms.security.procedures.unrestricted=apoc.*

Alternatives

You can download Neo4j Desktop and run locally for development, spin up a hosted Neo4j Sandbox instance, run Neo4j in one of the many cloud options, spin up Neo4j in a Docker container, on Arch linux you can install neo4j-community from AUR or on Debian-based systems install Neo4j from the Debian Repository. Just be sure to update the Neo4j connection string and credentials accordingly in backend/.env.

Start Neo4J and confirm the database is running at http://localhost:7474.

Operations on Neo4j 4.4

Docker or Docker Compose

• we need to set command: ["tail", "-f", "/dev/null"] in the Neo4j block of docker-compose.yml on top level so the Neo4j database is in maintenance mode

Create Neo4j Dump

To create a dump in Neo4j running in a Docker container:

• set the database to maintenance mode, see above
• entering the following commands:

# connect to the Docker containers Neo4j terminal
$ docker exec -it neo4j bash
# generate Dump
neo4j% neo4j-admin dump --to=/var/lib/neo4j/$(date +%F)-neo4j-dump
# exit bash
neo4j% exit
# copy the dump out of the running Docker container
$ docker cp <docker-image-name('neo4j')>:/var/lib/neo4j/neo4j-dump <local-folder-path>/$(date +%F)-neo4j-dump

If you need a specific database name, add the option --database=<name> to the command neo4j-admin dump.

In our deployments there are cases where the database is called neo4j (used by default) and in other cases graph.db (accidentally happened when we loaded the database into a new cluster).

Import Neo4j Dump

To import a dump into Neo4j running in a Docker container:

• set the database to maintenance mode, see above
• entering the following commands:

# copy the dump into the running Docker container
$ docker cp <local-folder-path>/neo4j-dump <docker-image-name('neo4j')>:/var/lib/neo4j/$(date +%F)-neo4j-dump
# connect to the Docker containers Neo4j terminal
$ docker exec -it neo4j bash
# to load the dump into the database we need the following command in this terminal
neo4j% neo4j-admin load --from /var/lib/neo4j/$(date +%F)-neo4j-dump --force
# leave the terminal by entering
neo4j% exit

If you need a specific database name, add the option --database=<name> to the command neo4j-admin load. To find out the default database name, see below.

Commands

Here we describe some rarely used Cypher commands for Neo4j that are needed from time to time:

Index And Constraint Commands

If indexes or constraints are missing or not set correctly, the browser search will not work or the database seed for development will not work.

The indexes and constraints of our database are set in backend/src/db/migrate/store.js. This is where the magic happens.

It's called by our prod:migrate init command. This command initializes the Admin user and creates all necessary indexes and constraints in the Neo4j database.

Calls in development

Locally without Docker:

# in backend folder
$ yarn prod:migrate init

Locally with Docker:

# in main folder
$ docker compose exec backend yarn prod:migrate init

Calls in production

Locally with Docker:

# in main folder
$ docker compose exec backend /bin/sh -c "yarn prod:migrate init"

On a server with Kubernetes cluster:

# tested for one backend replica
# !!! be aware of the kubectl context !!!
$ kubectl -n default exec -it $(kubectl -n default get pods | grep ocelot-backend | awk '{ print $1 }') -- /bin/sh -c "yarn prod:migrate init"

Enter and Exit Cypher Shell

# enter the bash of Neo4j database
$ kubectl -n default exec -it $(kubectl -n default get pods | grep ocelot-neo4j | awk '{ print $1 }') -- bash
# enter Cypher-Shell to send Cypher commands
neo4j# cypher-shell

# send Cypher commands
cypher-shell# < Cypher commands >
# exit Cypher-Shell
cypher-shell# :exit

# exit bash
neo4j# exit

Cypher commands to show indexes and constraints

# in browser command line or Cypher shell

# show all indexes and constraints
$ :schema

# show all indexes
$ CALL db.indexes();

# show all constraints
$ CALL db.constraints();

Cypher commands to create and drop indexes and constraints

# in browser command line or Cypher shell

# create indexes
$ CALL db.index.fulltext.createNodeIndex("post_fulltext_search",["Post"],["title", "content"]);
$ CALL db.index.fulltext.createNodeIndex("user_fulltext_search",["User"],["name", "slug"]);
$ CALL db.index.fulltext.createNodeIndex("tag_fulltext_search",["Tag"],["id"]);

# drop an index
$ DROP CONSTRAINT ON ( image:Image ) ASSERT image.url IS UNIQUE

# drop all indexes and constraints
$ CALL apoc.schema.assert({},{},true) YIELD label, key RETURN * ;

Database Management Commands

Cypher commands to manage databases

# in browser command line or Cypher shell

# show the default database
$ SHOW DEFAULT DATABASE
# show all databases
$ SHOW DATABASES

To set the default database by configuration, use NEO4J_dbms_default__database as an environment variable when starting Neo4j 4.4, see Docker specific configuration settings.

If a database with this name does not exist, an empty database with this name is created and all other databases remain. You can switch back to an existing database without damaging it.

It seems to be impossible to change the name of an existing database on the fly. To change the name of an existing database, we need to load a dump or restore a backup under a new name in Neo4j.

For more information on deployment, see Default Database Name.