Create job flows with parent-child dependencies.

The FlowProducer allows you to create complex job hierarchies where parent jobs wait for their children to complete before being processed.

All flow operations are atomic - either all jobs in a flow are added, or none are. This is achieved using Redis MULTI/EXEC transactions.

Usage

# Create a flow with parent and children
{:ok, flow} = BullMQ.FlowProducer.add(%{
  name: "process_order",
  queue_name: "orders",
  data: %{order_id: 123},
  children: [
    %{name: "validate", queue_name: "validation", data: %{order_id: 123}},
    %{name: "check_inventory", queue_name: "inventory", data: %{order_id: 123}},
    %{name: "process_payment", queue_name: "payments", data: %{order_id: 123}}
  ]
}, connection: :redis)

# The parent job will be processed only after all children complete

Flow Structure

A flow is defined as a tree of jobs:

%{
  name: "parent_job",
  queue_name: "main_queue",
  data: %{...},
  opts: %{...},
  children: [
    %{name: "child1", queue_name: "queue1", data: %{...}},
    %{
      name: "child2",
      queue_name: "queue2",
      data: %{...},
      children: [
        %{name: "grandchild", queue_name: "queue3", data: %{...}}
      ]
    }
  ]
}

Accessing Child Results

When a parent job is processed, use methods on BullMQ.Job to access children:

def process(%Job{} = job) do
  # Get children results
  {:ok, children_values} = BullMQ.Job.get_children_values(job)

  # Get ignored failures (when using ignore_dependency_on_failure)
  {:ok, ignored} = BullMQ.Job.get_ignored_children_failures(job)

  # Get pending dependencies
  {:ok, deps} = BullMQ.Job.get_dependencies(job)

  {:ok, process_with_children(job.data, children_values)}
end

Failure Handling

By default, if a child fails, the parent will also fail. You can control this behavior with options:

• :fail_parent_on_failure - If false, parent continues even if children fail
• :ignore_dependency_on_failure - Ignore failed child and continue