>Digital-flame

✅ Short Answer

You apply a Liquibase migration using the update command from the CLI:

liquibase update

This command tells Liquibase to:

• Connect to your database
• Read the changelog file
• Apply any changesets that haven’t been run yet

🔎 Step-by-Step Command Line Usage

1. ✅ Basic Example

liquibase \
  --changeLogFile=db/changelog.xml \
  --url=jdbc:postgresql://localhost:5432/mydb \
  --username=myuser \
  --password=mypass \
  update

This runs all unapplied changesets in changelog.xml.

2. ✅ Set Configuration Once (Optional)

Create a liquibase.properties file to avoid typing all parameters every time:

changeLogFile=db/changelog.xml
url=jdbc:postgresql://localhost:5432/mydb
username=myuser
password=mypass
driver=org.postgresql.Driver

Then you can run:

liquibase update

✅ Other Useful CLI Commands

🧠 Behind the scenes

• Liquibase checks the DATABASECHANGELOG table
• Compares changesets against what’s already been applied
• Runs only new changesets (unless runAlways or runOnChange is used)

📌 Key Takeaways

✅ Use the update command to apply migrations via the command line
✅ Prefer using a liquibase.properties file for reusability
✅ Useful for local dev, automation, or CI/CD pipelines
✅ Track progress and debug with commands like status, history, and updateSQL

✅ Short Answer

Yes, Liquibase can generate a changelog from an existing database schema using the generateChangeLog command.
This reverse-engineers the current database structure into a changelog file (XML, YAML, JSON, or SQL) so you can start tracking and versioning it with Liquibase.

🔎 How It Works

You run:

liquibase generateChangeLog \
  --url=jdbc:your_database_url \
  --username=your_user \
  --password=your_pass \
  --changeLogFile=initial-changelog.xml

🔹 Liquibase connects to your database
🔹 Reads the current schema (tables, columns, constraints, indexes, sequences…)
🔹 Writes them out as a Liquibase changelog

You can specify:

• --changeLogFile → path to write the changelog
• --format → output format: XML (default), YAML, JSON, or SQL

🧪 Example Output (XML)

<changeSet id="1" author="liquibase">
  <createTable tableName="users">
    <column name="id" type="BIGINT" autoIncrement="true"/>
    <column name="email" type="VARCHAR(255)"/>
  </createTable>
</changeSet>

⚠️ Things to Keep in Mind

• ✅ Great for bootstrapping Liquibase into an existing project
• ❌ Not ideal for ongoing migrations — generated changelogs may be large and unstructured
• ✅ After generation, you should clean up, organize, and commit to version control

📌 Key Takeaways

✅ Liquibase can generate a changelog from your existing DB schema using generateChangeLog
✅ This is ideal for initial setup or reverse-engineering legacy databases
✅ You can output in XML, YAML, JSON, or SQL formats and customize the scope

✅ Short Answer

Liquibase tracks applied changes by recording each successfully executed changeset in a special database table called DATABASECHANGELOG.
Each entry includes metadata like id, author, filename, checksum, and execution timestamp.

🔎 Detailed Explanation

When you run liquibase update, Liquibase:

1. Reads your changelog file(s)
2. For each <changeSet>:
   • Checks if it’s already listed in the DATABASECHANGELOG table
   • Compares the checksum to detect changes
3. Executes the changeset only if it hasn’t been run yet, or if you’ve specified runAlways or runOnChange
4. Records a new row in DATABASECHANGELOG after successful execution

📋 Key Columns in DATABASECHANGELOG

🧠 Why is this important?

• ✅ Ensures each changeset runs exactly once
• ✅ Supports safe retries — rerunning update won’t re-apply executed changesets
• ✅ Detects unauthorized or accidental modifications via checksum mismatch
• ✅ Enables rollback, tagging, auditing, and repeatable deployments

🔄 Related Concepts

• Checksum mismatch → means the changeset was edited after execution
• clearCheckSums command → resets checksums if you deliberately modified a changeset and want Liquibase to accept the new version
• runOnChange="true" → Liquibase re-applies the changeset if the checksum changes (use with caution)

📌 Key Takeaways

✅ Liquibase uses the DATABASECHANGELOG table to track exactly which changesets were applied, when, and in what order.
✅ This mechanism ensures idempotency, auditability, and integrity of schema changes across environments.
✅ You never need to manually track what’s been applied — Liquibase does it all.

✅ Short Answer

runOnChange="true" tells Liquibase to re-run a changeset if its definition changes, even if it was already applied.

By default, Liquibase only runs each changeset once, but runOnChange forces Liquibase to track changes via checksums and re-apply the changeset if it detects any modifications.

🔎 Detailed Explanation

🔸 Without runOnChange

Liquibase stores a checksum for each applied changeset in the DATABASECHANGELOG table.

If the changeset is later modified, Liquibase will:

• ❌ Detect a checksum mismatch
• ❌ Throw a validation error
• ✅ Require you to fix it manually or use clearCheckSums

This behavior prevents untracked schema drift.

🔸 With runOnChange="true"

If a changeset has:

<changeSet id="1" author="alice" runOnChange="true">
  <createView viewName="user_view" ... />
</changeSet>

Liquibase will:

• Detect a checksum change
• ✅ Re-execute the changeset automatically
• ✅ Update the DATABASECHANGELOG record with the new checksum

🧠 Common Use Cases

⚠️ Caution

Use runOnChange carefully:

• It re-applies changes each time the checksum changes — which can overwrite custom manual edits in prod
• Not recommended for large data operations (e.g. insert or update)

📌 Key Takeaways

✅ runOnChange="true" re-executes a changeset if it has been modified after execution
✅ Useful for views, procedures, and other objects that evolve over time
❌ Should be used with caution — especially in production environments

✅ Short Answer

Preconditions in Liquibase are conditions that must be true before a changeset or changelog is applied.
If the precondition fails, Liquibase can be told to halt, warn, or continue, depending on your configuration.

🔎 Detailed Explanation

A precondition acts like a safeguard — for example:

• “Only run this change if a table exists”
• “Skip this if a column is already present”
• “Apply this only on PostgreSQL”

You can define preconditions:

• At the changelog level (applies to all changesets)
• At the changeset level (applies to one changeset only)

✅ Common Use Cases for Preconditions

🧪 Example – XML format

<changeSet id="1" author="alice">
  <preConditions onFail="HALT">
    <not>
      <tableExists tableName="users"/>
    </not>
  </preConditions>

  <createTable tableName="users">
    <column name="id" type="int" autoIncrement="true"/>
    <column name="username" type="varchar(255)"/>
  </createTable>
</changeSet>

✅ This change will only run if the users table does not exist.
If the table does exist, Liquibase will halt the migration (because onFail="HALT").

🛠️ Other Precondition Types

• sqlCheck – Run custom SQL that returns true/false
• changeLogPropertyDefined
• columnExists
• sequenceExists
• foreignKeyConstraintExists
• customPrecondition – write your own in Java

📌 Key Takeaways

✅ Preconditions let you add logic before a changeset runs, preventing errors or unintended changes.
✅ You can halt, warn, or skip based on the result.
✅ They are useful for idempotency, environment targeting, and fail-safety.

✅ Short Answer

The DATABASECHANGELOG table is a system table created and maintained by Liquibase.
It tracks which changesets have already been executed, preventing duplicates and ensuring safe, repeatable database migrations.

🔎 Detailed Explanation

Whenever Liquibase applies a changeset from your changelog file, it logs a record in the DATABASECHANGELOG table. This allows Liquibase to:

• Skip already executed changesets
• Validate checksums (to detect tampering)
• Maintain the execution order
• Enable rollbacks and tagging

📋 Key Columns in DATABASECHANGELOG

🧠 Why Is This Table Important?

• ✅ Prevents running the same changes twice
• ✅ Ensures safe migrations across environments (dev, test, prod)
• ✅ Enables rollback and versioning
• ✅ Supports audit and compliance tracking

📌 Key Takeaways

✅ DATABASECHANGELOG is the core mechanism Liquibase uses to track schema changes.
✅ It ensures that each changeset runs exactly once, safely and in order.
✅ It is stored inside your target database, alongside your actual tables.

✅ Short Answer

Liquibase supports the following changelog file formats:

1. XML
2. YAML
3. JSON
4. Formatted SQL

🔎 Details

📝 Example file names:

• changelog.xml
• db-changelog.yaml
• changelog.json
• changelog.sql (with --liquibase formatted sql at the top)

📌 Key Takeaways

✅ Liquibase supports XML, YAML, JSON, and SQL as changelog formats.
✅ Choose based on team preference, tooling, and readability needs.
✅ All formats support the same core features like changesets, rollback, and tagging.

✅ Short Answer

Yes, database changes can be rolled back — but only under specific conditions:

• ✅ In a transaction → changes can be rolled back automatically on failure.
• ❌ Most schema changes (DDL) like CREATE TABLE are not transactional in many databases.
• ✅ Migration tools like Liquibase can rollback if you explicitly define rollback logic.
• ❌ Tools like Flyway do not rollback — you must undo changes manually.

🔎 1. Database Transaction Rollback

✅ If you’re performing DML operations (insert, update, delete) in a transaction:

BEGIN;
UPDATE users SET email = 'test@example.com' WHERE id = 1;
ROLLBACK; -- Undoes the update

✅ Works in most RDBMS (PostgreSQL, MySQL with InnoDB, etc.)

❌ But many DDL statements (like ALTER TABLE, CREATE TABLE) are auto-committed:

• PostgreSQL supports transactional DDL
• MySQL/MariaDB does not (DDL can’t be rolled back)
• Oracle supports transactional DDL in some cases
• SQLite does support it

🔎 2. Liquibase Rollback

✅ You can define explicit rollback in changesets:

<changeSet id="3" author="alice">
  <addColumn tableName="users">
    <column name="phone" type="varchar(20)"/>
  </addColumn>
  <rollback>
    <dropColumn tableName="users" columnName="phone"/>
  </rollback>
</changeSet>

Then run

liquibase rollbackCount 1

✅ Liquibase will run the rollback logic to revert changes.

❌ But this is manual — rollback only happens when you run it.
❌ There’s no automatic rollback on error during a migration.

🔎 3. Flyway Rollback

❌ Flyway has no rollback feature.

✅ If a migration fails:

• The script is marked as failed
• You must fix manually
• Optionally write an “undo” migration (U1__undo_users.sql), but Flyway won’t run it automatically

📌 Key Takeaways

✅ Short Answer

When a migration fails:

• The migration process stops immediately
• The failed changeset is not marked as executed
• Any partial changes already applied by that changeset remain in the database unless you handle rollback manually

❗ No automatic rollback is performed unless you explicitly define and run one.

🔎 Liquibase – What happens on failure?

🔹 Suppose you run:

liquibase update

🔹 Scenario:

One of your changesets (e.g. id=3) has a syntax error or constraint violation.

🔸 What happens:

• Liquibase stops at the error.
• That changeset is not marked as executed in DATABASECHANGELOG
• Previous changesets remain applied
• Partial SQL from the failed changeset may have affected the DB, depending on DB settings and the operation

✅ You can:

• Fix the issue
• Re-run liquibase update (it will skip changesets already marked as executed)

🔹 Example:

<changeSet id="1" author="alice">
  <createTable tableName="users">
    <column name="id" type="bigint" autoIncrement="true"/>
  </createTable>
</changeSet>

<changeSet id="2" author="alice">
  <addColumn tableName="users">
    <column name="email" type="varchar(255)"/>
  </addColumn>
</changeSet>

<changeSet id="3" author="alice">
  <!-- This fails due to bad SQL -->
  <sql>ALTER TABLE users ADD age_in_days_typo INT</sql>
</changeSet>

✅ Changeset 3 fails
✅ Changesets 1 and 2 are marked as executed
❌ 3 is not
❗ You must fix the SQL, then re-run

🔹 Rollback?

Liquibase does not rollback automatically on failure.
You can:

• Define a <rollback> section per changeset
• Manually run:

liquibase rollbackCount 1

But rollback must be explicitly triggered, not automatic.

🔎 Flyway – What happens on failure?

In Flyway:

• The failed script is recorded with a failure status in flyway_schema_history
• The entire migration run is stopped
• You must repair or fix the problem before continuing

✅ Fix the file or write a new script
✅ Then run:

flyway repair  # marks the failed migration as pending or removes bad checksum
flyway migrate

Flyway does not support rollback, so you must clean the database (dev only) or manually fix the schema.

📌 Key Takeaways

✅ Short Answer

Migrations are tracked using a special table in your database where the tool records every changeset or script it has applied — so it knows what’s already been done and what’s still pending.

🔎 Liquibase Tracking

Liquibase uses two special tables:

✅ When you run liquibase update, it checks DATABASECHANGELOG to see:

• Which id + author combinations have already been run
• Whether the file’s checksum changed (to detect unauthorized edits)

Each applied changeset gets one row in DATABASECHANGELOG.

✅ Columns in DATABASECHANGELOG include:

• ID — the unique changeset ID
• AUTHOR — who authored the change
• FILENAME — changelog file
• DATEEXECUTED — timestamp
• MD5SUM — to detect changes
• ORDEREXECUTED — order of application
• TAG — optional release label

🔎 Flyway Tracking

Flyway uses a single table called:

✅ It records:

• Version (e.g., 1, 2.1)
• Description (from filename)
• Script name (e.g., V1__create_users.sql)
• Checksum for file integrity
• Success flag (false if the migration failed)

🔒 Locking & Safety

Both tools:

• Use a locking mechanism to prevent concurrent migrations
• Check checksums → if someone modifies a file after it’s been applied, you get a validation error (unless you force it)

📌 Key Takeaways

✅ Migration tools record applied changes in a special DB table (DATABASECHANGELOG or flyway_schema_history)
✅ This ensures:

• Migrations are applied once
• No changes are skipped
• Modified scripts trigger checksum errors

✅ These tables are part of the DB itself — portable, visible, and auditable.