https://github.com/artefactual/archivematica/issues/938 introduces a new architecture where workers process file-level tasks in batches. When an instance of MCPClient is allocated one of these batches it looks up whether the client module prefers to be provisioned multiple times (as processes) to process the batch in smaller chunks. The module claims this behaviour by defining a concurrent_instances function that returns the number of instances required.
Currently, this is only used in the following cases:
https://github.com/artefactual/archivematica/issues/938 introduces a new architecture where workers process file-level tasks in batches. When an instance of MCPClient is allocated one of these batches it looks up whether the client module prefers to be provisioned multiple times (as processes) to process the batch in smaller chunks. The module claims this behaviour by defining a
concurrent_instances
function that returns the number of instances required.Currently, this is only used in the following cases:
In all the cases above the integer returned comes from
multiprocessing.cpu_count()
.We may want to revise this in order to:
concurrent_instances
in other modules.