Workers

Helix Swarm uses background processes, called workers, to respond to events in the Helix Core Server. The default number of workers is three, and each worker processes events for up to 10 minutes. When a worker terminates, a new one is spawned.

Note

Each worker maintains a connection to the Helix Core Server for the duration of its lifetime. This may impact your Helix Core Server management practices.

Tip

To view Swarm task queue information, as an admin or super user, navigate to the User id dropdown menu, select System Information, and open the Queue Info tab.

Worker status

To determine the current status of workers, do one of the following:

  • As an admin or super user, navigate to the User id dropdown menu, select System Information, and open the Queue Info tab.

  • or

  • Visit the URL: https://myswarm.url/queue/status.

  • The response is formatted in JSON, and looks like this:

    {"tasks":0,"futureTasks":1,"workers":3,"maxWorkers":3,"workerLifetime":"595s"}

During normal use of Swarm, the following error message appears for logged-in users when Swarm detects that no workers are running:

Worker configuration

Tip

If you make a configuration change, Swarm will not use it until the configuration cache has been reloaded, this forces Swarm to use the new configuration. You must be an admin or super user to reload the Swarm config cache. Navigate to the User id dropdown menu, select System Information, click the Cache Info tab, and click the Reload Configuration button.

To adjust the configuration for workers, add a configuration block to the SWARM_ROOT/data/config.php file:

<?php
// this block should be a peer of 'p4'
'queue' => array(
'workers' => 3, // defaults to 3
'worker_lifetime' => 595, // defaults to 10 minutes (less 5 seconds)
'worker_task_timeout' => 1800, // defaults to 30 minutes
'worker_memory_limit' => '1G', // defaults to 1 gigabyte
),

where:

  • workers specifies the number of worker processes that should be available. The default is 3. The cron job ensures that new worker processes are started when necessary. If the limit is reached or exceeded, new worker processes are not started.
  • worker_lifetime specifies the amount of time in seconds that a worker process should run for. The default is 595 seconds (10 minutes less 5 seconds). If a worker process exceeds this limit while processing a task, it will complete the active task and then terminate. worker_lifetime does not cause tasks to terminate mid-processing.
  • worker_task_timeout specifies the maximum amount of time in seconds that a worker process can spend processing a single task. The default is 1800 seconds (30 minutes). This is useful for terminating workers that might get stalled in a variety of situations.
  • worker_memory_limit specifies the maximum amount of memory that a worker process is allowed to use while processing a task. The default is 1G (1 gigabyte).

Manually start workers

To kick off a new worker process, do one of the following:

  • As an admin or super user, navigate to the User id dropdown menu, select System Information, open the Queue Info tab, and click Start a worker.

  • or

  • Visit the URL: https://myswarm.url/queue/worker.

If the number of workers running matches the configured limit, the requested worker process is not started.

Note

This technique does start a worker, but it lasts only for its configured lifetime. Typically, you would always want at least one worker running. See Set up a recurring task to spawn workers for details.

Manually restart workers

To restart an idle worker process, remove its lock file:

rm data/queue/workers/worker_id

A worker process that is busy processing a task will continue operation until its task is complete. Immediately afterwards, if the worker notices that its lock file is missing it exits.

If you have a recurring task to start workers, the recurring task starts a fresh worker, if necessary. See Set up a recurring task to spawn workers for details.