multiple-vms.md

MULTIPLE VMS

Activate uses memory efficiently by maintaining soft references for the entities that are loaded from the storage. To use this approach in an environment with multiple virtual machines using the same storage, there is the Optimistic Offline Locking.

This is a widely known pattern to deal with concurrency. It uses an entity version information to detect conflicts using the underlying storage.

The propose of offline locking is to detect transaction conflicts at commit time (optimistic locking), then Activate can reload the entity information and retry the transaction if it is possible.

For a detailed explanation of the offline locking internals, please refer to the Architecture Documentation.

CONFIGURATION

There are two system properties to configure the offline locking:

• activate.offlineLocking.enable

  If this parameter is true, the offline locking is enabled. Default: false.

• activate.offlineLocking.validateReads

  If this parameter is true, the offline locking also validate all transactions reads against the storage version information. Default: false.

IMPORTANT: It is necessary to create the version column for all entities. It must be a Long value. Example migration:

class ExampleMigration extends Migration {
	def timestamp = 201208191834l
	def up = {
		table[MyEntity]
			.addColumn(
				_.column[Long]("version"))
	}
}

IMPORTANT: The read validation mechanism doesn't validate reads against the database if only immutable values are used (vals).

DEFERRED READ VALIDATION

The deffered read validation allows to postpone the read validation.

For each entity:

import myPersistenceContext._
import scala.concurrent.duration._

class MyEntity extends Entity {
	
	override def shouldValidateRead =
		deferFor(5 minutes)
}

For any entity:

import net.fwbrasil.activate.ActivateContext
import scala.concurrent.duration._

class myPersistenceContext extends ActivateContext {

	val storage = ...
	
	override def shouldValidateRead(entity: BaseEntity) =
		deferFor(5 minutes)
}