Phraseanet integrates a worker manager in order to perform asynchronous work. This manager will soon replace Task Manager.
Each worker asynchronously is assigned for the following types of tasks:
Thus each type of task to be performed, we can assign one or more workers as needed.
Workers use RabbitMQ, a message broker to operate.
Worker-specific commands are integrated into the command line utility Konsole Commander:
worker
worker:execute Listen queues define on configuration, launch
corresponding service for execution
worker:run-service Execute a service
worker:show-configuration Show queues configuration
bin/console worker:execute -m 10
bin/console worker:execute --queue-name=recordedit -m 10
bin/console worker:execute --queue-name=recordedit --queue-name=ftp -m 10
bin/console worker:execute --queue-name=assetsIngest --queue-name=createRecord
--queue-name=deleteRecord --queue-name=exportMail --queue-name=exposeUpload
--queue-name=ftp --queue-name=mainQueue --queue-name=populateIndex
--queue-name=pullAssets --queue-name=subtitle --queue-name=validationReminder
--queue-name=webhook -m 3
To access the “Worker manager” panel:
This section allows you to define for each type of task the maximum number of attempts as well as the time between each attempt.
Note
For certain tasks (example: “Subdef creation”), a third parameter is used to set the interval (in seconds) of attempted execution for deferred messages.
Note
For any change of configuration on a queue, it will have to be destroyed via the Queues tab for the modifications to be taken into account.
This section is used to display the status of the jobs in the form of a list. Each job has four possible states:
Warning
Les travaux en erreurs publiés dans les queues dead-letter dédiées aux erreurs (nommées “xxx_failed”) sont automatiquement effacées de la file RabbitMQ au bout de 96 heures. Jobs in error published in the dead-letter error queues (named “xxx_failed”) are automatically deleted from the RabbitMQ queue after 96 hours.
See below dead-letter queues examples dedicated to errors (framed in red):
Allows you to manually restart indexing on one or more databases.