pgroll

pgroll is an open-source command-line tool for teams changing PostgreSQL schemas while applications are still rolling out. It suits engineers who need reversible, low-disruption migrations and are comfortable operating database tooling themselves. Its strongest case is coordinated application and schema change; PostgreSQL 14 users relying on row-level security should choose another approach.

Overview

pgroll uses an expand/contract workflow so old and new schema versions remain usable at the same time. Client applications choose a version by setting PostgreSQL’s search_path when they connect, allowing an application rollout to move independently from the database change.

The tool is free and open source, with the repository licensed under Apache License 2.0. It runs as a single cross-platform binary and can be self-hosted. Questions and issue reports go through GitHub Issues, GitHub Discussions, or Discord rather than a commercial support desk.

Key features

Reversible, low-disruption migrations

pgroll describes its migrations as zero-downtime and applies changes without locking the database. An active migration can be rolled back to the previous schema version, which gives teams a practical escape route when a new application release misbehaves. Automatic column backfills reduce routine migration work when a change requires existing rows to receive values.

Virtual schemas and write propagation

For breaking changes, pgroll builds virtual schemas with views over physical tables and uses triggers to propagate writes between old and new columns. That design preserves compatibility during the transition, but it also introduces database objects and behavior that operators must understand before relying on them in production.

PostgreSQL and ORM coverage

pgroll supports PostgreSQL 14 and later and says it works with any PostgreSQL service, including Amazon RDS and Aurora. It can convert SQL migration statements produced by ORMs, with examples for Alembic, Django, and Drizzle. ORM output is not a hands-off path: the project’s guide requires manual up and down data migrations, and generated SQL may need to be consolidated into fewer pgroll operations.

Important PostgreSQL 14 caveat

On PostgreSQL 14, row-level security policies are not respected by pgroll’s versioned views. Teams using RLS on that release should treat pgroll as a poor fit, regardless of the appeal of reversible migrations.

Pricing

There are no paid tiers, seat charges, trial periods, or renewal terms in this plan. The trade-off is operational: you run the tool yourself and rely on the project’s community channels for help. It fits teams that can manage PostgreSQL migration infrastructure and want Apache-licensed software without a subscription.

Platforms

pgroll supports Linux, macOS, Windows, and self-hosted deployments. Its stated database target is PostgreSQL 14 and later, including managed services such as RDS and Aurora.

Who it's for

Choose pgroll when a PostgreSQL-backed application needs schema changes staged across releases, especially when rollback and simultaneous schema versions matter. It is well suited to engineering teams that control deployment and connection settings and can review SQL generated by their ORM.

Look elsewhere if your PostgreSQL 14 deployment depends on row-level security, or if you want migrations fully generated and maintained by an ORM without manual data steps and SQL consolidation. The self-hosted, command-line model also favors teams comfortable owning operations rather than buyers seeking a managed service.

Pros and cons

• Pros: Reversible active migrations provide a direct recovery path during application rollout.
• Pros: Expand/contract versioning keeps old and new clients working concurrently.
• Pros: Automatic column backfills and trigger-based write propagation address common breaking-change mechanics.
• Pros: Free Apache 2.0 software runs across major desktop platforms and self-hosted environments.
• Cons: PostgreSQL 14 versioned views do not respect row-level security policies.
• Cons: ORM-generated migrations still require manual up and down data migrations and possible SQL consolidation.
• Cons: Operating a command-line, self-hosted tool leaves deployment and support responsibility with your team.

Alternatives

For a broader directory of migration choices, see Database Migration Tools; for validation and testing-oriented products, see Database Testing Tools.

• Amazon Deequ is a free Apache 2.0 library for data quality checks, but it requires Apache Spark and a matching release.
• DataKitchen TestGen offers a free forever open-source plan with one user, one connection, and one project, making it a better fit for bounded test environments.
• Atlas is an open-core migration tool whose core engine is Apache 2.0 licensed, with advanced features in proprietary tiers.
• Great Expectations focuses on data expectations; its Developer plan is free for up to five data assets under test monthly and three users.
• HammerDB is free, open-source database benchmarking software for Windows and Linux, not a schema rollout workflow.
• Soda provides freemium pipeline testing, metrics observability, and alerting rather than pgroll’s versioned migration mechanism.
• Sqitch is a free database change-management application when you want a permissively usable migration tool without pgroll’s virtual-schema approach.
• SQLancer is a free MIT-licensed tool for database testing, suited to finding database bugs rather than coordinating application schema versions.

Verdict

pgroll is a strong choice for PostgreSQL teams that need reversible, zero-downtime-style schema rollouts and can operate a self-hosted command-line tool. Its main reason to choose it is simultaneous old/new schema compatibility with rollback; its main reason to look elsewhere is the PostgreSQL 14 row-level-security failure and the manual work left around ORM migrations.