tech-docs

Technical documentation for ArchivesSpace

View the Project on GitHub archivesspace/tech-docs

Edit this page on GitHub architecture/jobs/README.md

Report issue on Jira architecture/jobs/README.md

Background Jobs

ArchivesSpace provides a mechanism for long running processes to run asynchronously. These processes are called Background Jobs.

Managing Jobs in the Staff UI

The Create menu has a Background Job option which shows a submenu of job types that the current user has permission to create. (See below for more information about job permissions and hidden jobs.) Selecting one of these options will take the user to a form to enter any parameters required for the job and then to create it.

When a job is created it is placed in the Background Job Queue. Jobs in the queue will be run in the order they were created. (See below for more information about multiple threads and concurrent jobs.)

The Browse menu has a Background Jobs option. This takes the user to a list of jobs arranged by their status. The user can then view the details of a job, and cancel it if they have permission.

Permissions

A user must have the create_job permission to create a job. By default, this permission is included in the repository_basic_data_entry group.

A user must have the cancel_job permission to cancel a job. By default, this permission is included in the repository_managers group.

When a JobRunner registers it can specify additional create and cancel permissions. (See below for more information)

Types, Runners and Schemas

Each job has a type, and each type has a registered runner to run jobs of that type and JSONModel schema to define its parameters.

Registered JobRunners

All jobs of a type are handled by a registered JobRunner. The job runner classes are located here:

  backend/app/lib/job_runners/

It is possible to define additional job runners from a plugin. (See below for more information about plugins.)

A job runner class must subclass JobRunner, register to run one or more job types, and implement a #run method for jobs that it handles.

When a job runner registers for a job type, it can set some options:

For more information about defining a job runner, see the JobRunner superclass:

  backend/app/lib/job_runner.rb

JSONModel Schemas

A job type also requires a JSONModel schema that defines the parameters to run a job of the type. The schema name must be the same as the type that the runner registers for. For example:

  common/schemas/import_job.rb

This schema, for JSONModel(:import_job), defines the parameters for running a job of type import_job.

Concurrency

ArchivesSpace can be configured to run more than one background job at a time. By default, there will be two threads available to run background jobs. The configuration looks like this:

  AppConfig[:job_thread_count] = 2

The BackgroundJobQueue will start this number of threads at start up. Those threads will then poll for queued jobs and run them.

When a job runner registers, it can set an option called :run_concurrently. This is false by default. When set to false a job thread will not run a job if there is already a job of that type running. The job will remain on the queue and will be run when there are no longer any jobs of its type running.

When set to true a job will be run when it comes to the front of the queue regardless of whether there is a job of the same type running.

Plugins

It is possible to add a new job type from a plugin. ArchivesSpace includes a plugin that demonstrates how to do this:

  plugins/jobs_example