pgEdge Docker

This repository contains the Dockerfile used to build pgedge/pgedge on Docker Hub.

See the example commands below for running pgEdge containers in Docker. You will need to provide a JSON configuration file that specifies the database nodes and users.

Examples

Single Node

To run a single node you can use this command:

docker run -it --rm -p 5432:5432 \
  -v "./examples/singlenode/db.json:/home/pgedge/db.json" \
  pgedge/pgedge

You can then log in using psql with the following command:

PGPASSWORD=uFR44yr69C4mZa72g3JQ37GX \
psql -h localhost -p 5432 -U admin defaultdb

And of course you should customize the user passwords before using this in any real deployment.

Multi-Node

A Docker Swarm example of a two node cluster is located at examples/swarm.

Data Volumes

This image is compatible with Docker volumes and bind mounts. The configuration for both is similar. Because PostgreSQL requires the data directory to be owned by the user running the database, the PGDATA directory should be specified as a subdirectory of the volume mount.

By default, this image uses the following approach for volume configuration:

• /data is the volume mount point
• /data/pgdata is the PostgreSQL data directory (PGDATA)

An example Docker compose spec that bind mounts the host folder ./n1 to the container's /data folder looks like this:

postgres-n1:
  image: pgedge/pgedge:latest
  environment:
    - 'NODE_NAME=n1'
    - 'PGDATA=/data/pgdata'
  volumes:
    - './db.json:/home/pgedge/db.json'
    - './n1:/data'

You can also take a look at examples/swarm/stack.yaml for an example of using Docker volumes.

Automatic DDL Replication

Automatically replicating DDL statements is available on an opt-in basis. To enable this behavior, set the autoddl:enabled option on the database. Currently this option must be set at the time of database creation. Changing the option after creation is possible, but is not yet documented here.

{
  "options": ["autoddl:enabled"]
}

See the full example JSON configuration in the Database Configuration section below.

The user running DDL statements must be a superuser currently in order to use automatic DDL replication.

More information on automatic DDL replication can be found here.

Database Configuration

A simple JSON file is used to configure the database nodes. You can customize this according to your needs, including adding more nodes and users. Always remember to change the passwords!

{
  "name": "defaultdb",
  "port": 5432,
  "options": ["autoddl:enabled"],
  "nodes": [
    {
      "name": "n1",
      "region": "us-east-1",
      "hostname": "postgres-n1"
    },
    {
      "name": "n2",
      "region": "us-east-2",
      "hostname": "postgres-n2"
    }
  ],
  "users": [
    {
      "username": "admin",
      "password": "uFR44yr69C4mZa72g3JQ37GX",
      "superuser": true,
      "service": "postgres",
      "type": "admin"
    },
    {
      "username": "pgedge",
      "password": "z1Zsku10a91RS526jnVrLC39",
      "superuser": true,
      "service": "postgres",
      "type": "internal_admin"
    },
    {
      "username": "pgcat_auth",
      "password": "5Y306TW24540dEnyxp3mQBwH",
      "service": "postgres",
      "type": "pooler_auth"
    },
    {
      "username": "pgcat_admin",
      "password": "k6uu4od8r0P6lA11Oep648KC",
      "service": "pgcat",
      "type": "other"
    }
  ]
}

Spock Notes

When the container runs, the Spock extension is created and replication subscriptions are automatically created.

If you do not use automatic DDL replication as described above, you can instead use the spock.replicate_ddl function to manually replicate DDL statements. For example:

SELECT spock.replicate_ddl('CREATE TABLE public.users (id uuid, name text, PRIMARY KEY (id))');

The users table will now exist on all nodes. Now you can add the table to the default replication set by executing this command on all nodes:

SELECT spock.repset_add_all_tables('default', ARRAY['public']);