Laravel Dynamic Servers Laravel Package

title: How servers are managed under the hood weight: 2

To use the package, you don't need to know how things are working on the hood. However, to troubleshoot things, or when you create your own server provider, it might be handy to know what is going on. On this page, we'll describe how the package works internally.

An important thing to know is that the package tracks state of all servers it has created in the dynamic_servers table.

Creating and starting servers

Whenever you call DynamicServers::increase() (or ::ensure() and it creates servers for you), we will create a new Spatie\DynamicServers\Models\Server model. When it is created it will have the status new.

Right after such a model is created, we will call start() on it.

// happens internally

$server->start():

This will start a job, Spatie\DynamicServers\Jobs\StopServer that will call the startServer method on the ServerProvider class of the server. The Server model will now have status starting.

Most hosting servers cannot start servers instantly. It takes a while to provision them. That's why the CreateServerJob will dispatch another job, VerifyServerStartedJob to verify that a server has been started. The VerifyServerStartedJob will use the hasStarted method on the ServerProvider class to determine if the server has been fully started at the hosting provider. If the server has not started, the VerifyServerStartedJob will release it self, and another attempt will be made 20 seconds later.

If hasStarted returns true, then VerifyServerStartedJob will finish properly and the Server model will be marked as running.

Stopping and deleting servers

Whenever you call DynamicServers::decrease() (or ::ensure() and it removes servers for you), we will call stop() on those servers.

// happens internally

$server->stop():

This will start a job, Spatie\DynamicServers\Jobs\StopServerJob that will call the stopServer method on the ServerProvider class of the server. The Server model will now have status stopping.

Most hosting servers cannot stop servers instantly. It takes a while to shut them down them. That's why the StopServerJob will dispatch another job, VerifyServerStoppedJob to verify that a server has been stopped. The VerifyServerStoppedJob will use the hasStopped method on the ServerProvider class to determine if the server has been fully stopped by the hosting provider. If the server has not stopped, the VerifyServerStoppedJob will release itself, and another attempt will be made 20 seconds later.

If hasStopped returns true, then VerifyServerStoppedJob will mark Server model stopped. Next, the DeleteServerJob will be called. This job will call deleteServer on the ServerProvider to delete the server. The Server model will have status deleting

The VerifyServerDeletedJob will be dispatched to verify that a server has been deleting. The VerifyServerDeletedJob will use the hasBeenDeleted method on the ServerProvider class to determine if the server has been fully stopped by the hosting provider. If the server has not stopped, the VerifyServerDeletedJob will release itself, and another attempt will be made 20 seconds later.

If hasBeenDeleted returns true, then the Server will have status deleted.