Workers¶
Missing Dependency Packages¶
If any required dependency packages for a task are missing, the workflow does not proceed and a message like Worker missing […] is shown. To install the missing packages:
Use the top-level menu .
Right-click the worker to open the context-menu and click .
Select the indicated missing packages.
When the installation finishes the task will automatically start processing.
This requires administrator permissions.
What are workers?¶
XamFlow supports processing workflow tasks on multiple computers. These computers are called workers. By default one local worker is running on the current computer where XamFlow was installed, but additional workers can be added to process workflows on a cluster of computers by installing XamFlow Worker Service on these computers.
The XamFlow system can be run on one local computer in isolation. Or multiple users can connect to a central local self-hosted XamFlow Server, to allow concurrent collaboration on workflows, and sharing of data, workflows and computer resources.
See Worker Configuration for more information about configuring a worker.
What are dependency packages?¶
XamFlow allows managing additional software to be installed on worker computers via Dependency Packages. Initially only the most commonly used packages are installed automatically. Additional, more specialized packages can be installed when required.
Concurrent Processing¶
Processing task jobs are running on workers. By default each worker allows processing one task at a time. Multiple worker computers can be connected. But even with only one worker, each worker can also be configured to process multiple jobs concurrently.
Use the top-level menu .
Right-click the worker to open the context-menu and click .
Set the maximum number of jobs.
This requires administrator permissions. Carefully monitor the computer resource utilization of CPU, memory (RAM), GPU, GPU memory, and disks, for example in Windows Task Manager’s Performance tab. If any resource approaches 100%, consider reducing the maximum number of concurrent jobs, to avoid out-of-memory errors and overall impact on performance and throughput.
