Database Diff Tool

Overview

Compare database schemas between two environments (development vs. staging, staging vs.

Prerequisites

• Connection credentials to both source and target databases
• psql or mysql CLI configured to connect to both environments
• Read access to informationschema and pgcatalog (PostgreSQL) or information_schema (MySQL)
• Permission to run pg_dump --schema-only for full schema extraction
• Understanding of which environment is the "source of truth" (typically the migration-managed environment)

Instructions

1. Extract the full schema from both databases for comparison:

• PostgreSQL: pgdump --schema-only --no-owner --no-privileges -f schemasource.sql sourcedb and repeat for targetdb
• MySQL: mysqldump --no-data --routines --triggers sourcedb > schemasource.sql
• Alternatively, query information_schema directly for programmatic comparison

1. Compare tables present in each database:

• SELECT tablename FROM informationschema.tables WHERE tableschema = 'public' AND tablecatalog = 'sourcedb' EXCEPT SELECT tablename FROM informationschema.tables WHERE tableschema = 'public' AND tablecatalog = 'targetdb'
• This reveals tables that exist in source but not in target (and vice versa)

1. Compare columns for each shared table:

• Query informationschema.columns from both databases for: columnname, datatype, charactermaximumlength, isnullable, columndefault, ordinalposition
• Flag differences in data type, nullability, default values, and column ordering
• Detect added columns (in source, not target) and dropped columns (in target, not source)

1. Compare indexes:

• PostgreSQL: Query pg_indexes for indexname, indexdef on each database
• MySQL: Query informationschema.STATISTICS for INDEXNAME, COLUMNNAME, NONUNIQUE
• Flag missing, extra, or differently-defined indexes

1. Compare constraints (primary keys, foreign keys, unique, check):

• Query informationschema.tableconstraints and informationschema.keycolumn_usage
• Detect missing foreign keys, changed constraint names, and altered check constraint expressions

1. Compare functions, stored procedures, and triggers:

• PostgreSQL: Query pgproc for function signatures and pgtrigger for trigger definitions
• MySQL: Query informationschema.ROUTINES and informationschema.TRIGGERS
• Compare function bodies for logical differences

1. Compare enum types and custom types (PostgreSQL):

• Query pgtype and pgenum for enum label differences
• Detect added or removed enum values (note: PostgreSQL only supports adding enum values, not removing)

1. Generate a structured diff report categorizing differences as:

• Added: Objects in source not present in target (require CREATE statements)
• Removed: Objects in target not present in source (require DROP statements, confirm intentional)
• Modified: Objects differing between source and target (require ALTER statements)

1. Generate migration SQL to synchronize the target database to match the source:

• CREATE TABLE for new tables, ALTER TABLE ADD COLUMN for new columns
• ALTER TABLE ALTER COLUMN for type changes, ALTER TABLE DROP COLUMN for removed columns
• CREATE INDEX / DROP INDEX for index differences
• Include transaction wrapping and rollback-safe operations

1. Validate the generated migration by applying it to a copy of the target database and re-running the diff. The second diff should report zero differences, confirming the migration produces the expected state.

Output

• Schema diff report listing all differences categorized by type (added, removed, modified)
• Migration SQL script to synchronize target schema to match source
• Rollback SQL script to reverse the migration if needed
• Side-by-side comparison of differing object definitions
• Drift detection summary highlighting changes not tracked in migration files

Error Handling

Examples

Detecting schema drift between staging and production: After 3 months without auditing, the diff reveals: 2 columns added to production manually (not in migrations), 1 index missing from staging, and 3 functions with different implementations. A migration script is generated to bring staging in sync, and the manual production changes are backported into migration files.

Pre-deployment schema validation: Before deploying a release with 5 migration files, run the diff between the post-migration staging schema and the expected schema. The diff catches a migration that accidentally dropped a constraint that a later migration depends on. The migration ordering is fixed before production deployment.

Comparing PostgreSQL schemas across major version upgrade: Schema extracted from PostgreSQL 14 and compared against PostgreSQL 16 after migration. Diff reveals function signature changes for built-in function calls, updated default values for new parameters, and deprecated syntax in stored procedures. Migration script updates function definitions for the new version.

Resources

• PostgreSQL information_schema: https://www.postgresql.org/docs/current/information-schema.html
• MySQL information_schema: https://dev.mysql.com/doc/refman/8.0/en/information-schema.html
• pg_dump schema-only mode: https://www.postgresql.org/docs/current/app-pgdump.html
• migra (PostgreSQL schema diff tool): https://github.com/djrobstep/migra
• Skeema (MySQL schema management): https://www.skeema.io/