cql-track 1.1.0

Schema migration tool for Apache Cassandra

CQLTrack - A Cassandra Based Schema Versioning Tool

Schema migration tool for Apache Cassandra.

Version-controlled .cql files, distributed locking, checksum validation, multi-environment profiles, static analysis, and schema diffing — built on the DataStax Python driver.

Compatibility: Verified with Apache Cassandra (3.x / 4.x / 5.x) and DataStax Enterprise / Astra DB. AWS Keyspaces is not supported at this time due to differences in LWT, schema agreement, and authentication mechanisms.

Installation

PyPI (recommended)

pip install cql-track

CLI

Once installed, the cqltrack command is available globally:

cqltrack --help

Docker

# start local Cassandra
docker compose up -d

# run cqltrack in a container
docker build -t cqltrack .
docker run --rm -v $(pwd)/migrations:/workspace/migrations \
  -v $(pwd)/cqltrack.yml:/workspace/cqltrack.yml \
  --network host cqltrack migrate

From source

git clone https://github.com/ereshzealous/cql-track.git
cd cql-track
pip install -e .

Why cqltrack?

Most migration tools target relational databases. Cassandra is different — DDL is asynchronous, there are no transactions, and clusters span multiple datacenters. cqltrack is purpose-built for these realities:

• Schema agreement handling — waits for all nodes to converge between DDL statements
• Distributed locking — Lightweight Transactions (LWT) prevent concurrent migration runs across CI workers or deploy nodes
• Partial failure tracking — if statement 3 of 5 fails, cqltrack records the failure and lets you fix and retry cleanly
• Cassandra-native — supports NetworkTopologyStrategy, all consistency levels, LWT serial consistency for locks, and Astra DB

Features

Migrate & Rollback	Versioned .cql files with -- @down rollback sections
Distributed Lock	LWT-based lock with configurable TTL and auto-expiry
Checksum Validation	MD5 checksums detect modified migration files
Multi-Environment	YAML profiles for dev/staging/prod in one config file
Static Analysis	Linter catches dangerous patterns before they hit your cluster
Schema Diff	Compare schemas across keyspaces or environments
Schema Snapshot	Export live schema as CQL
Partial Failure	Failed migrations are tracked and retryable
Baseline	Adopt cqltrack on existing databases without re-running history
CI/CD Ready	pending command as a deploy gate, JSON output, exit codes
SSL/TLS	Full TLS support including mutual TLS with client certificates
Astra DB	Secure connect bundle support for DataStax Astra
Programmatic API	Use in FastAPI, Django, or plain Python at app startup

Quick Start

# install
pip install cql-track

# start local Cassandra (optional)
docker compose up -d

# create config
cat > cqltrack.yml <<EOF
cassandra:
  contact_points: [127.0.0.1]
  port: 9042
keyspace:
  name: my_app
  replication:
    class: SimpleStrategy
    replication_factor: 1
migrations:
  directory: migrations
EOF

# initialize keyspace and tracking tables
cqltrack init

# create a migration
cqltrack new create_users_table

Edit migrations/V001__create_users_table.cql:

CREATE TABLE IF NOT EXISTS users (
    user_id    UUID,
    email      text,
    name       text,
    created_at timestamp,
    PRIMARY KEY (user_id)
);

-- @down
DROP TABLE IF EXISTS users;

# apply
cqltrack migrate

# check status
cqltrack status

Programmatic API

Use cqltrack as a library in your Python applications:

from cqltrack import CqlTrack

# with YAML config
with CqlTrack("cqltrack.yml") as t:
    t.init()
    t.migrate()
    print(t.status())

# inline config (no YAML file needed)
with CqlTrack(contact_points=["127.0.0.1"], keyspace="my_app") as t:
    t.migrate()

See examples/ for FastAPI, Django, and plain Python integration guides.

Commands

cqltrack init        Create the keyspace and tracking tables
cqltrack migrate     Apply pending migrations
cqltrack rollback    Undo the most recently applied migration(s)
cqltrack status      Show applied and pending migrations
cqltrack history     Full migration history with timing and who applied
cqltrack pending     Exit code 1 if unapplied migrations exist (CI gate)
cqltrack baseline    Mark migrations as applied without executing (adoption)
cqltrack validate    Check that applied files haven't been modified
cqltrack repair      Update recorded checksums to match files on disk
cqltrack new         Scaffold a new migration file
cqltrack lint        Static analysis for dangerous CQL patterns
cqltrack snapshot    Export the live keyspace schema as CQL
cqltrack diff        Compare schemas between environments or keyspaces
cqltrack profiles    List available environment profiles

Global options:

-c, --config     Path to cqltrack.yml
--profile        Environment profile (dev, staging, prod)
-k, --keyspace   Override target keyspace
--host           Cassandra contact point(s), comma-separated
-p, --port       Cassandra native transport port
--json           Machine-readable JSON output

Configuration

cassandra:
  contact_points: [127.0.0.1]
  port: 9042
  auth:
    username: myuser
    password: mypass
  consistency: LOCAL_QUORUM       # default: LOCAL_ONE
  connect_timeout: 10
  request_timeout: 30
  ssl:
    enabled: true
    ca_certs: /path/to/ca.crt
    verify: true
  # secure_connect_bundle: /path/to/astra-bundle.zip

keyspace:
  name: my_app
  replication:
    class: NetworkTopologyStrategy
    dc1: 3
    dc2: 2

migrations:
  directory: migrations

lock:
  ttl: 600

schema:
  agreement_wait: 30

profiles:
  dev:
    keyspace:
      name: myapp_dev
  prod:
    cassandra:
      contact_points: [10.0.2.1, 10.0.2.2, 10.0.2.3]
      consistency: LOCAL_QUORUM
    keyspace:
      name: myapp_prod

Config resolution order (last wins):

1. Built-in defaults
2. YAML file values
3. Profile overrides (--profile prod)
4. Environment variables (CQLTRACK_KEYSPACE, CQLTRACK_PASSWORD, etc.)
5. CLI flags (--keyspace, --host, --port)

Migration File Format

Files follow the naming convention V<version>__<description>.cql:

migrations/
  V001__create_users_table.cql
  V002__create_orders_table.cql
  V003__add_phone_to_users.cql

Each file has an UP section and an optional -- @down section for rollback:

-- UP: applied by `cqltrack migrate`
ALTER TABLE users ADD phone text;

-- @down
-- applied by `cqltrack rollback`
ALTER TABLE users DROP phone;

Linter

The built-in linter catches common mistakes before they hit your cluster:

$ cqltrack lint
  ERR   V005:3   [drop-no-if-exists]       DROP without IF EXISTS — not idempotent
  WARN  V008:1   [create-no-if-not-exists]  CREATE without IF NOT EXISTS
  ERR   V012:5   [column-drop]             ALTER TABLE DROP — permanently deletes data

Rules: no-rollback, empty-rollback, drop-no-if-exists, create-no-if-not-exists, column-drop, truncate, pk-alter, type-change

CI/CD

Use pending as a deploy gate:

cqltrack --profile prod pending || exit 1

Use lint in pre-commit or PR checks:

cqltrack lint

JSON output for tooling:

cqltrack --json status
cqltrack --json history
cqltrack --json pending
cqltrack --json lint
cqltrack --json diff --target-keyspace other_ks

Requirements

• Python 3.9+
• Apache Cassandra 3.x / 4.x / 5.x, or DataStax Enterprise / Astra DB
• Dependencies: cassandra-driver, click, pyyaml

Examples

See the examples/ directory for integration guides:

• CLI usage — full command reference with examples
• Plain Python — programmatic API, context managers, error handling
• FastAPI — lifespan pattern, migration status endpoint
• Django — AppConfig.ready(), management commands

Sample migration files are in examples/migrations/.

Documentation

See USAGE.md for the complete guide — configuration reference, all commands with examples, SSL/TLS setup, Astra DB, distributed locking internals, partial failure handling, schema agreement, adoption workflow, lint rules, and troubleshooting.

Contributing

See CONTRIBUTING.md for development setup, testing, and how to submit changes.

License

MIT