Init PRO#

The init command connects to an existing database, inspects its schema, and prints a complete config to stdout. The output includes globals, up, seed, deseed, and down sections ready to use with the other commands.

Flags#

mongodb, cassandra, and redis have no relational catalog to introspect, so init rejects them with init does not support driver "redis", use scaffold instead. Use scaffold for those, which builds a config from a table and column description rather than a live database.

edg init \
--driver pgx \
--url "postgres://root@localhost:26257/defaultdb?sslmode=disable" \
--schema public > workload.edg

Generated Config#

The generated config includes:

• up CREATE TABLE statements derived from the database’s own DDL (CockroachDB’s SHOW CREATE TABLE, MySQL’s SHOW CREATE TABLE, Oracle’s DBMS_METADATA.GET_DDL, reconstructed from sys catalog views for MSSQL, or reconstructed from INFORMATION_SCHEMA for Spanner, or read straight out of sqlite_master for SQLite).
• seed One INSERT per table with an expression for each non-generated column. Columns with auto-increment, identity, or default functions like gen_random_uuid() and now() are skipped. Expressions are chosen by data type (e.g. uuid_v4() for UUID, uniform.int(1, 1000) for INT, gen('sentence:3') for VARCHAR). CHECK BETWEEN constraints are detected and used to narrow the range.
• deseed TRUNCATE (pgx, oracle) or DELETE FROM (mysql, mssql, spanner, sqlite) in reverse dependency order.
• down DROP TABLE in reverse dependency order.
• Tables are topologically sorted so parent tables are created before children.

The output is a starting point. You’ll typically want to refine the seed expressions to produce more realistic data, add a run section, and adjust globals.rows.

Examples by driver#

CockroachDB / PostgreSQL (pgx)

edg init \
--driver pgx \
--url "postgres://root@localhost:26257/dbname?sslmode=disable" \
--schema public > workload.edg

Aurora DSQL

edg init \
--driver dsql \
--url "clusterid.dsql.us-east-1.on.aws" \
--schema public > workload.edg

MySQL

edg init \
--driver mysql \
--url "root:password@tcp(localhost:3306)/dbname?parseTime=true" \
--database dbname > workload.edg

MSSQL

edg init \
--driver mssql \
--url "sqlserver://sa:P4ssw0rd@localhost:1433?database=master&encrypt=disable" \
--schema dbo > workload.edg

Oracle

edg init \
--driver oracle \
--url "oracle://system:password@localhost:1521/dbname" \
--schema SYSTEM > workload.edg

SQLite

The database name is the attached schema, which is main unless you’ve ATTACHed others.

edg init \
--driver sqlite \
--url "file:edg.db" \
--database main > workload.edg

Google Cloud Spanner (GoogleSQL)

edg init \
--driver spanner \
--url "projects/my-project/instances/my-instance/databases/my-db" \
--license "$EDG_LICENSE" \
--schema "" > workload.edg