Using the Appkit Tasks Module
The Task module lets you run numerous, time-consuming tasks (atomic units of work) behind the scenes, while continuing to use your application. While your tasks are running, you can monitor their status as well as stop, restart, and cancel them if needed.
Setup
Updating the pom.xml
To add the Task module to a Maven project, add this dependency to your project’s pom.xml
file:
<dependency>
<groupId>twigkit</groupId>
<artifactId>twigkit.task</artifactId>
<version>${project.parent.version}</version>
</dependency>
Configuring the task executor
The Task module includes its own task executor that takes charge of managing your tasks. Under the hood, the executor uses a registry that can be configured in order to fine-tune how your tasks will be handled.
To customize the configuration of the task executor, first create a file named executor.conf
and place this in src/main/resources/conf/services/task/
. Within that file, these parameters can be set:
max-tasks-in-memory
: the maximum number of tasks the registry can hold in memory at any one time. The default is 100.
max-tasks-on-disk
: the maximum number of tasks that can be persisted on disk. The default is 10000000.
disk-store-location
: the location for storing/persisting tasks. The default is java.io.tmpdir
- a JVM system property that will change depending on your Operating System.
time-to-idle
: the amount of time in seconds a task is allowed to be left idle (from last access or modified date) before being removed from the registry. The default is 3600.
time-to-live
: the amount of time in seconds a task is allowed to exist (from its creation date) before being removed from the registry regardless of how often it is used. The default is effectively forever.
overflow-to-disk
: whether additional tasks will be stored on disk, if the in-memory registry is full. The default is true.
persist-to-disk
: whether tasks will be persisted to disk, for example, if the if in-memory registry is full or the application shuts down. The default is true.
As noted, all of the above parameters already contain default values that will be used in case executor.conf
cannot be found or one or other of the parameters is not defined.
Writing your own tasks
To submit a task to the task executor, the actual actions of the task (what it will do when it starts, stops, and is deleted) must first be defined. A task is written in Java and must follow a prescribed template. This template is as follows:
@XmlRootElement
public class MyTask extends Task {
public MyTask() {
this(null);
}
public MyTask(User owner) {
super(owner);
}
public MyTask(User owner, Map<String, String> attributes) {
super(owner, attributes);
}
@Override
public void start() throws TaskException {
// Put here what the task will do when started, for example, download a file
}
@Override
public void start(Callback<?> callback) throws TaskException {
// This is a work in progress; Do not use.
}
@Override
public void stop() throws TaskException {
// Put here what the task will do when stopped
}
@Override
public void delete() throws TaskException {
// Put here what the task will do when deleted
}
}
As can be seen from the template, three methods are available that can be used to decide what the task will do when it is started (start
), when it is stopped (stop
), and when it is deleted (delete
) from the registry. These methods are named when particular RESTful endpoints are hit as discussed below.
Configuring your own tasks
After you have written your task (see above), the next step is to create a configuration file for that task. This will let you submit the task to the Task Web Service and define specific properties of the task that will be available to the task when it is started, stopped etc.
As with any other configuration, the task configuration must be stored in resources/conf
, for example, resources/conf/tasks/download.conf
. The simplest configuration for a task would be one with just the name
parameter:
name: package1.package2.package3.MyTask
Where name
is the fully-qualified class name of the Java class you have written to define your Task
.
The configuration can also include a display name, as well as any other attributes that the task might need, for example:
display: Download
destination: /downloads
file-type: pdf
zipped: true
These additional attributes can be accessed within the task via getAttributes()
.
Usage
Working with tasks via the Appkit Task RESTful endpoint
The Task web service provides a number of RESTful endpoints that can be used to submit, query, and cancel your tasks. These endpoints are as follows:
GET /twigkit/api/tasks
: this will get a list of tasks owned by the current user.
GET /twigkit/api/tasks/{id}
: this will get any information on the task with the given ID, including status and any task-specific attributes.
GET /twigkit/api/tasks/{id}/status
this will get the status of the task with the given ID.
POST /twigkit/api/tasks/{task}
: this will submit a task with the given configuration task name to the task executor. The given task name is the configuration name for a task. For example, if the configuration for a particular task was stored in resources/conf/tasks/download.conf
, the task name passed to the submit endpoint would be tasks.download
.
POST /twigkit/api/tasks/{id}/stop
: this will stop the task with the given ID.
POST /twigkit/api/tasks/{id}/restart
: this will restart the task with the given ID
DELETE /twigkit/api/tasks/{id}
: this will cancel the task with the given ID. This will call stop
and delete
on the task as well as remove the task from the registry.
Task persistence
By default, while an application is running, tasks are periodically flushed to disk. For this purpose, in the disk-store-location
(see above) a task-cache.data
file is created to keep track of tasks until the application shuts down when an additional file,task-cache.index
, is created to persist tasks. Next time the application is run these tasks will be available for continued use.
To disable task persistence, set the configuration parameters overflow-to-disk
and persist-to-disk
to false
.
Task expiration
By default, tasks will not be stored in the registry indefinitely. To customize how long tasks may be stored, two configuration options are available:
-
time-to-live
This is the amount of time in seconds a task is allowed to exist before being removed from the registry regardless of how often it is used. The default is effectively forever. This is useful in case resources are limited and it is not clear whether tasks will be used after a finite amount of time.
-
time-to-idle
This is the amount of time in seconds a task is allowed to be left idle before being removed from the registry. The default is one hour.
Whatever times are set for these values, they are never reset. This holds true even when the application is not running. If the application is shut down and enough time elapses to allow the task to expire, on restart, the task will no longer be available for use.
To disable task expiration, set time-to-live
and time-to-idle
to 0.
Deleting a task after expiration
After a task has expired, it is not only deleted from the registry, but the delete
method implemented within your task will be named. If there is any final action that you would like your task to perform before it is removed then you can implement that action in the tasks delete
method.
Task eviction
By default, tasks are stored on disk if the capacity to store tasks in-memory is exceeded. In this case, if the number of tasks being stored exceeds the limit set by the configuration parameter max-tasks-on-disk
then the least recently used task will be evicted from the registry. The default maximum number of tasks that can be stored on disk is 10000000.
If instead you decide not to store tasks on disk (both overflow-to-disk
and persist-to-disk
are false
) but would like to limit the number of tasks that can be stored in-memory, then the configuration parameter max-tasks-in-memory
can be set. This will evict the least recently used task once capacity is reached.
Deleting a task after eviction
After a task has been evicted, it is not only deleted from the registry, but the delete
method implemented within your task will be named. If there is any final action that you would like your task to perform before it is removed then you can implement that action in the tasks delete
method.