Module: Legion::Extensions::Llm::Openai::Runners::FleetWorker

Extended by:
Logging::Helper
Defined in:
lib/legion/extensions/llm/openai/runners/fleet_worker.rb

Overview

Runner entrypoint for OpenAI fleet request execution. Subscription dispatch invokes this via Legion::Runner.run as runner_class.send('handle_fleet_request', **envelope) — the fleet envelope arrives as kwargs (symbol keys), never as a positional payload.

Class Method Summary collapse

Class Method Details

.handle_fleet_request(**envelope) ⇒ Object 



21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
 
# File 'lib/legion/extensions/llm/openai/runners/fleet_worker.rb', line 21

def handle_fleet_request(**envelope)
  log.debug do
    "Handling OpenAI fleet request: request_id=#{envelope[:request_id] || 'unknown'}, " \
      "provider_instance=#{envelope[:provider_instance] || 'unknown'}"
  end

  Legion::Extensions::Llm::Fleet::ProviderResponder.call(
    payload: envelope,
    provider_family: Openai::PROVIDER_FAMILY,
    provider_class: Openai::Provider,
    provider_instances: -> { Openai.discover_instances }
  )
rescue StandardError => e
  handle_exception(e, level: :error, handled: true, operation: 'openai.fleet_worker.handle_request')
  raise
end