When you launch the worker node application it displays its main window, which summarises the current status of the worker and has commands to let you edit the configuration, display the detailed status of a particular task, and view the log file.
On the Mac the worker node continues to run in the background even when the main window is closed. Under Windows it terminates when you close the window.
Three distinct processes run concurrently and independently in the worker node, the File Watcher (which periodically scans the watch folder(s) looking for new files), the Server Watcher (which performs queries against the server looking for clips that are flagged for processing), and the Worker itself (which fetches tasks from the task list and performs them). The main window has check boxes which you can use to enable or disable these processes and also displays what they are currently doing.
At the heart of the worker node operation is the task list, a combined queue of tasks waiting to be performed and list of tasks that have already completed. Each task represents a clip or file that needs to be processed and a job to be applied to it.
The main window shows your current tasks in a table, with one task per row, whether completed or queued waiting to run. Each task has details of the input file and the job to be applied to it, together with information such as when the status was last updated, when the file was last seen (for example, if it the file is on a removable or network volume and is currently offline), or whether the task entry was created by the current running instance of the worker node or not (either because the software has been restarted or because another instance on another machine is using the same work file).
A task can be in one of several different states:
- Queued: When a new file appears in one of the watch folders a new task is created and queued. When the worker runs it will start executing the first available task whose status is ‘queued’.
- Running: This means the task is currently being executed. Note that several worker nodes on different machines could be working on the same task list, so it’s not necessarily being executed by the instance on this machine. (In practice though, if the instance is not the current one the most likely explanation is that an earlier instance was terminated prematurely and the worker node software has been restarted.)
- Failed: The task has completed but an error occurred. See the Information field for further details. If you correct the problem (eg. by starting the workgroup server, if it was previously down) you can resubmit the task so it is performed again.
- Complete: The task has successfully completed
- Hold: If you want to temporarily defer tasks (for example, to allow tasks that were queued later to be executed first) you can mark them as being on hold.
- Offline: This status is similar to tasks being on Hold. Tasks that were previously queued are automatically set to Offline when the worker node starts up if the file they relate to is no longer available (for example, because a server volume is temporarily offline).
- Skipped: Some jobs (such as importing an image sequence or a P2 metaclip) process multiple files in one go. If a file was queued for processing but has already been processed by another job it is marked as Skipped.
You can resubmit failed tasks (for example, after restarting the CatDV server), hold or resume tasks, or delete tasks using the relevant buttons at the bottom of the task list.
Tasks have a priority which affect the order in which they are done. The job definition defines a default priority but you can adjust the priority of specific tasks in the queue if you want to bring one forward.
Tasks are normally hidden after 3 days (unless you check the Show All box) and deleted altogether after 14 days.