Managing databases with Maestro
About deploying databases
We currently support the following databases, with no need for additional configuration after deployment.
- MySQL (or Percona if configured via Manifest)
- SQLite (only in development environments)
When creating a Maestro application, you can add as many databases as you need in your service configuration during the application build.
After you have analyzed your code, ensure that your desired database type is displayed in the About your app section of the analysis results.
When we deploy a database we automatically generate the required users and passwords to allow authentication. You can find these values via your Dashboard in the detail page of any database server.
They will be available as environment variables and your application will be configured to use them.
Connecting your app to your DB in Maestro
Databases in Maestro run as separate components and aren’t containerized. Even though they may be running on the same private network as your cluster servers, you will not be able to connect to them via localhost because of the nature of containers (which are, by definition, abstracted from operating the system).
Instead, we we automatically create a Kubernetes services for each DB that connects out from your cluster to the database server(s). To see these on your cluster, you can list the namespaces and then select your namespace and list the services associated with it with the following commands:
$ kubectl get namespaces $ kubectl -n <your-namespace> get svc
This will show you all the services running, some of which will be your databases.
Service names for database groups
If your application has two or more database groups, your Kubernetes database services will inherit those names. For example, if you have three MySQL database groups named
archive then the Kubernetes services will be named:
If one of these groups is set as your “primary” then it will use the default service name (
mysql) instead of its group-specific name.
Connection strings in Maestro
A typical connection string might have:
- The protocol
- The username and password (where required)
- The name of the Kubernetes service, including database groups (e.g.
- The name of the DB server (e.g.
So to connect to a MongoDB server named
mongo_production_1 and running in your app namespace as
mongodb-spare you would use something like:
A similar setup for a Postgres server would look something like this:
Connecting your app to your DB in Maestro V1
To connect to a database in Version 1 of Maestro, you should use its ElasticDNS address. Please read our full guide on ElasticDNS for more details.
Database deployment types
No database (external)
This option allows you to deploy your application without a database managed by Cloud 66, and is ideal for externally hosted databases.
You can also configure an external database via your Manifest file by specifying the
server node as
Please note that if there is no connectivity to your external database, or your external database host is not configured correctly, the deployment will fail.
This option deploys your chosen database to the same server as your web server - this is intended primarily for development, as running your database locally in production is not advised. In this case, your application database configuration will be amended to target your local database server. If you scale up your web server, these settings will also be amend automatically to reflect your database configuration.
This option will automatically create a new server for your database and configure your application accordingly.
Upgrading your database
Cloud 66 will not do in-place database upgrades, because this process may cause your application to stop working or may not be possible automatically. To upgrade your database through Cloud 66, we recommend that you create a new application (at which point Cloud 66 will deploy the newer database version).
Once the new application is created, you can migrate data from your old application to your new application.
Customize your database configuration
You can customize the database configuration on your servers using CustomConfig. CustomConfig is available for MySQL, PostgreSQL, Redis and MongoDB.
Editing and committing your database CustomConfig will perform the following steps on every database server in your application, one by one, sequentially:
- Check your template for Liquid syntax errors
- Determine the correct server configuration and prepare general variables
- Prepare custom variables for your database type (e.g. server_state)
- Compile the database configuration based on the information from the server and database type
- Upload the configuration to the server
- Restart your database
🚨 A bad database configuration might stop your database from working. Take extra care to make sure the configuration is correct.
Database customization variables
There are a number of variables available for use in your database CustomConfig. Some are general for all database types, while others are database specific.
The following variables are available to any database CustomConfig.
|server||Hash||Hash containing information about your server|
|memory||integer||Server memory size (bytes)|
|core||integer||Server core count|
The following variables are only available in the MySQL CustomConfig.
|server_state||string||Value can be stand_alone, mysql_master or mysql_slave based on your server status|
|server_id||integer||An ID used by MySQL replication to identify your server*|
*It is 0 for standalone servers, 1 for master servers and a number greater than 1 for slave servers
The following variables are only available in the PostgreSQL CustomConfig.
|server_state||string||Value can be stand_alone, pg_master or pg_slave based on your server status|
The following variables are only available in the Redis CustomConfig.
|server_state||string||Value can be stand_alone, redis_master or redis_slave based on your server status|
|master_address||string||IP address of replication master (empty string if server is stand alone or master)|
|master_port||integer||Will be 6379 when server is redis_slave , otherwise it is 0|