Skip to content
Sync data from one Postgres database to another
Branch: master
Clone or download

Latest commit


Type Name Latest commit message Commit time
Failed to load latest commit information.
exe Updated slop Mar 31, 2020
lib Added SQL conditions to non-tty output Mar 31, 2020
test Test unknown table Mar 29, 2020
.gitignore Added .pgsync.yml to gitignore Apr 5, 2016
.travis.yml Updated Travis to bionic Mar 19, 2020 Fixed behavior of wildcard without schema Mar 30, 2020
Gemfile First commit Dec 8, 2015
LICENSE.txt Bundler.with_clean_env is deprecated [skip ci] Mar 28, 2020
Rakefile Hide warnings when running tests Aug 15, 2017
config.yml Better init code Mar 28, 2020
pgsync.gemspec Updated slop Mar 31, 2020


Sync data from one Postgres database to another (like pg_dump/pg_restore). Designed for:

  • speed - tables are transferred in parallel
  • security - built-in methods to prevent sensitive data from ever leaving the server
  • flexibility - gracefully handles schema differences, like missing columns and extra columns
  • convenience - sync partial tables, groups of tables, and related records

🍊 Battle-tested at Instacart

Build Status


pgsync is a command line tool. To install, run:

gem install pgsync

This will give you the pgsync command. If installation fails, you may need to install dependencies.


In your project directory, run:

pgsync --init

This creates .pgsync.yml for you to customize. We recommend checking this into your version control (assuming it doesn’t contain sensitive information). pgsync commands can be run from this directory or any subdirectory.

How to Use

Sync all tables


Note: pgsync assumes your schema is setup in your to database. See the schema section if that’s not the case.

Sync specific tables

pgsync table1,table2

Sync specific rows (existing rows are overwritten)

pgsync products "where store_id = 1"

You can also preserve existing rows

pgsync products "where store_id = 1" --preserve

Or truncate them

pgsync products "where store_id = 1" --truncate

Exclude Tables

pgsync --exclude users

To always exclude, add to .pgsync.yml.

  - table1
  - table2

For Rails, you probably want to exclude schema migrations and ActiveRecord metadata.

  - schema_migrations
  - ar_internal_metadata


Define groups in .pgsync.yml:

    - table1
    - table2

And run:

pgsync group1

You can also use groups to sync a specific record and associated records in other tables.

To get product 123 with its reviews, last 10 coupons, and store, use:

    products: "where id = {1}"
    reviews: "where product_id = {1}"
    coupons: "where product_id = {1} order by created_at desc limit 10"
    stores: "where id in (select store_id from products where id = {1})"

And run:

pgsync product:123


Note: pgsync is designed to sync data. You should use a schema migration tool to manage schema changes. The methods in this section are provided for convenience but not recommended.

Sync schema before the data

pgsync --schema-first

Note: This wipes out existing data

Specify tables

pgsync table1,table2 --schema-first

Or just the schema

pgsync --schema-only

pgsync does not try to sync Postgres extensions.

Data Protection

Always make sure your connection is secure when connecting to a database over a network you don’t fully trust. Your best option is to connect over SSH or a VPN. Another option is to use sslmode=verify-full. If you don’t do this, your database credentials can be compromised.

Sensitive Information

Prevent sensitive information like email addresses from leaving the remote server.

Define rules in .pgsync.yml:

  email: unique_email
  last_name: random_letter
  birthday: random_date
    value: secret
    statement: "(RANDOM() * 10)::int"
  encrypted_*: null

last_name matches all columns named last_name and users.last_name matches only the users table. Wildcards are supported, and the first matching rule is applied.

Options for replacement are:

  • unique_email
  • unique_phone
  • unique_secret
  • random_letter
  • random_int
  • random_date
  • random_time
  • random_ip
  • value
  • statement
  • null
  • untouched

Rules starting with unique_ require the table to have a primary key. unique_phone requires a numeric primary key.

Multiple Databases

To use with multiple databases, run:

pgsync --init db2

This creates .pgsync-db2.yml for you to edit. Specify a database in commands with:

pgsync --db db2


To keep you from accidentally overwriting production, the destination is limited to localhost or by default.

To use another host, add to_safe: true to your .pgsync.yml.

Large Tables

For extremely large tables, sync in batches.

pgsync large_table --in-batches

The script will resume where it left off when run again, making it great for backfills.

Foreign Keys

By default, tables are copied in parallel. If you use foreign keys, this can cause violations. You can specify tables to be copied serially with:

pgsync group1 --debug



pgsync --help


pgsync --version


Use groups when possible to take advantage of parallelism.

For Ruby scripts, you may need to do:

Bundler.with_unbundled_env do
  system "pgsync ..."


If installation fails, your system may be missing Ruby or libpq.

On Mac, run:

brew install postgresql

On Ubuntu, run:

sudo apt-get install ruby-dev libpq-dev build-essential



gem install pgsync

To use master, run:

gem install specific_install
gem specific_install

Related Projects

Also check out:

  • Dexter - The automatic indexer for Postgres
  • PgHero - A performance dashboard for Postgres
  • pgslice - Postgres partitioning as easy as pie


Inspired by heroku-pg-transfer.


Everyone is encouraged to help improve this project. Here are a few ways you can help:

To get started with development:

git clone
cd pgsync
bundle install

createdb pgsync_test1
createdb pgsync_test2
createdb pgsync_test3

bundle exec rake test
You can’t perform that action at this time.