spock

Overview

works on pgedge kernel fork

Version

Build

You can build the RPM / DEB packages for spock using pig build:

pig build pkg spock         # build RPM / DEB packages

Install

You can install spock directly. First, make sure the PGDG and PIGSTY repositories are added and enabled:

pig repo add pgsql -u          # Add repo and update cache

Install the extension using pig or apt/yum/dnf:

• Install
• pig
• dnf
• apt

pig install spock;          # Install for current active PG version

pig ext install -y spock -v 18  # PG 18
pig ext install -y spock -v 17  # PG 17
pig ext install -y spock -v 16  # PG 16
pig ext install -y spock -v 15  # PG 15

dnf install -y pgedge-18       # PG 18
dnf install -y pgedge-17       # PG 17
dnf install -y pgedge-16       # PG 16
dnf install -y pgedge-15       # PG 15

apt install -y pgedge-18   # PG 18
apt install -y pgedge-17   # PG 17
apt install -y pgedge-16   # PG 16
apt install -y pgedge-15   # PG 15

Preload:

shared_preload_libraries = 'spock';

Create Extension:

CREATE EXTENSION spock;

Usage

Sources:

• Spock v5.0.10 README
• Getting started
• Configuration reference
• Limitations
• Release notes

spock provides active-active logical replication for PostgreSQL 15 through 18. Each participating database is a Spock node; a multi-master topology is formed by creating directed subscriptions between nodes.

Configuration

In postgresql.conf:

wal_level = 'logical'
max_worker_processes = 10
max_replication_slots = 10
max_wal_senders = 10
shared_preload_libraries = 'spock'
track_commit_timestamp = on
spock.enable_ddl_replication = on
spock.include_ddl_repset = on

Enabling

CREATE EXTENSION spock;

Creating Nodes

On each node, create a node identity:

-- Node 1
SELECT spock.node_create(
    node_name := 'n1',
    dsn := 'host=10.0.0.5 port=5432 dbname=mydb'
);

-- Node 2
SELECT spock.node_create(
    node_name := 'n2',
    dsn := 'host=10.0.0.7 port=5432 dbname=mydb'
);

Creating Subscriptions

For multi-master, each node subscribes to every other node:

-- On n1: subscribe to n2
SELECT spock.sub_create(
    subscription_name := 'sub_n1n2',
    provider_dsn := 'host=10.0.0.7 port=5432 dbname=mydb'
);

-- On n2: subscribe to n1
SELECT spock.sub_create(
    subscription_name := 'sub_n2n1',
    provider_dsn := 'host=10.0.0.5 port=5432 dbname=mydb'
);

Replication Set Management

-- Add table to replication
SELECT spock.repset_add_table('default', 'my_table');

-- Remove table from replication
SELECT spock.repset_remove_table('default', 'my_table');

-- Add all tables in a schema
SELECT spock.repset_add_all_tables('default', '{public}');

Key Features

• Multi-master (active-active) replication
• Automatic DDL replication
• Conflict detection and resolution using commit timestamps
• Row and column filtering
• Supports PostgreSQL 15, 16, 17, and 18
• Tables must have primary keys and matching schemas across nodes

Operations and Caveats

• Install spock and add it to shared_preload_libraries on every participating server before creating nodes or subscriptions.
• Keep table definitions, data types, primary keys, and relevant unique indexes identical across nodes. Coordinate DDL even when DDL replication is enabled.
• Replicated tables need a primary key or another usable replica identity. Temporary and unlogged tables are not replication targets.
• Spock operates per database. Repeat extension and topology setup for each database that participates.
• Active-active conflict handling depends on commit timestamps and policy. Test simultaneous inserts and updates, especially nullable unique keys, before production use.
• Upstream documents platform/build requirements in the README; verify that the PostgreSQL build and Spock package used on every node are compatible.

Version 5.0.10

5.0.10 is a patch release in the 5.0 line. Its release notes include fixes for unique indexes containing NULL, NULLS NOT DISTINCT conflict handling, refreshing cached index metadata after an index is dropped, exception-path memory handling, and numerical version checks used during rolling patch upgrades. Upgrade every node to a compatible patch level and validate subscriptions after the rolling change.