Utilities to ease the maintenance of MySQL or MariaDB database versioning using a Git repository.
This project is just to bring an idea to life and is still in an early stage of development, so any help is welcomed! Feel free to contribute.
Required libraries to build with node-gyp.
# apt install libkrb5-dev libssl-dev
It's recommended to install the package globally.
# npm install -g @verdnatura/myt
$ myt <command>
You can also install locally and use the npx command to execute it.
$ npm install @verdnatura/myt
$ npx myt <command>
Execute myt with the desired command.
$ [npx] myt [-w|--workspace <string>] [-r|--remote <string>] [-d|--debug]
[-h|--help] <command> [<args>]
The default workspace directory is the current working directory and unless otherwise indicated, the default remote is local.
Database versioning commands:
Local server management commands:
Each command can have its own specific commandline options.
First of all you have to initialize the workspace.
$ myt init
Now you can configure Myt using myt.config.yml file, located at the root of your workspace. This file should include the project codename and schemas/tables which are exported when you use pull or dump commands.
Don't forget to initialize git (if it isn't initialized yet).
$ git init
Create database connection configuration for each environment at remotes folder using standard MySQL ini configuration files. The convention remote names are local, production and test.
remotes/[remote].ini
Once the basic configuration is done, routines can be imported from the database into the project, it is recommended to use the production remote.
$ myt pull production
From now on, you can use the project as if it were a standard git repository (since it is). To apply changes to the database run the push command on the desired remote.
$ myt push [<remote>] [--commit]
Routines are placed inside routines folder. All objects that have PL/SQL code are considered routines. It includes events, functions, procedures, triggers and views with the following structure.
routines
`- schema
|- events
| `- eventName.sql
|- functions
| `- functionName.sql
|- procedures
| `- procedureName.sql
|- triggers
| `- triggerName.sql
`- views
`- viewName.sql
Versions are placed inside versions folder with the following structure. Don't place your PL/SQL objects here, use the routines folder!
versions
|- 00001-firstVersionCodeName
| |- 00-firstExecutedScript.sql
| |- 01-secondScript.sql
| `- 99-lastScript.sql
`- 00002-secondVersion
|- 00-firstExecutedScript.sql
`- 00-sameNumbers.sql
We have introduced a new feature that allows users to apply version-specific changes based on their configured environment, or "realm". This ensures that certain changes are only applied when the user is operating within a specific realm, providing an additional layer of customization and control.
To make use of this feature, you need to configure your realm in the versionConfig
table. Set your realm by inserting or updating a record in this table. The realm
field should contain the identifier of your environment.
To designate a file as realm-specific, include a .your_realm.
segment before the file extension. For example, if your realm is set to 'ab', the file should be named like this: filename.ab.sql
.
Set your Realm: Configure your realm in the versionConfig
table.
Add Files: Place your realm-specific files in the versions
folder. Make sure to follow the naming convention.
Version Order: Files are applied in the existing version order. In addition to that, the realm is validated.
Apply Changes: Run the usual versioning commands. The realm-specific files will only be applied if your configured realm matches the realm in the file name.
If no realm is configured, realm-specific files will be ignored.
If you have a realm configured but the realm-specific files belong to a different realm, those files will also be ignored.
This feature allows for greater flexibility when working in different environments, making it easier to manage realm-specific changes in a shared repository.
The local server is created as a MariaDB Docker container using the base dump created with the dump command plus pushing local versions and changed routines.
You can create your local fixture and structure files.
Initializes an empty workspace.
$ myt init
Incorporates database routine changes into workspace.
$ myt pull [remote] [-f|--force] [-c|--checkout] [-u|--update] [-s|--sums]
When --checkout option is provided, it does the following before export:
Applies versions and routine changes into database.
$ myt push [<remote>] [-f|--force] [-c|--commit] [-s|--sums] [-t|--triggers]
Commit is saved into database only if --commit option is provided, it prevents from accidentally saving local commits into shared servers, causing subsequent pushes from other clients to fail because they can't get that commit from the git tree in order to get differences.
Creates a new version folder, when name is not specified it generates a random name mixing a color with a plant name.
$ myt version [<name>]
Creates a new routine file with a default template.
$ myt create [-t <type>] <schema>.<name>
Cleans all already applied versions older than maxOldVersions.
$ myt clean [-p|--purge]
Exports database structure and fixtures from remote into hidden files located in dump folder. If no remote is specified production is used.
$ myt dump [<remote>] [-l|--lock] [-t|--triggers]
Exports local database fixtures into dump/fixtures.sql files. If no remote is specified local is used.
$ myt fixtures [<remote>]
Builds and starts local database server container. It only rebuilds the image when dump have been modified.
$ myt run [-c|--ci] [-r|--random] [-t|--tmpfs] [-n|--network <string>]
Starts local database server container. It does the minium effort, if it doesn't exists calls the run command, if it is started does nothing. Keep in mind that when you do not rebuild the docker you may be using an outdated version of it.
$ myt start
The main reason for starting this project is because there are no fully free and open source migration tools available that allow versioning database routines with a standard CVS system as if they were normal application code.
Also, the existing tools are too complex and require too much knowledge to initiate a small project.