fix: remove PENDING_NO_WORKERS status

src/aap_eda/core/enums.py

@@ -71,7 +71,6 @@ class ActivationStatus(DjangoStrEnum):
  STARTING = "starting"
  RUNNING = "running"
  PENDING = "pending"
- PENDING_WORKERS_OFFLINE = "pending (workers offline)"
  FAILED = "failed"
  STOPPING = "stopping"
  STOPPED = "stopped"
@@ -104,9 +103,6 @@ class DefaultCredentialType(DjangoStrEnum):
 # TODO: rename to "RulebookProcessStatus" or "ParentProcessStatus"
 ACTIVATION_STATUS_MESSAGE_MAP = {
  ActivationStatus.PENDING: "Wait for a worker to be available to start activation", # noqa: E501
- ActivationStatus.PENDING_WORKERS_OFFLINE: "Wait for a worker to be available to" # noqa: E501
- " start activation (all workers in the" # noqa: E501
- " node are offline)", # noqa: E501
  ActivationStatus.STARTING: "Worker is starting activation",
  ActivationStatus.RUNNING: "Container running activation",
  ActivationStatus.STOPPING: "Activation is being disabled",

src/aap_eda/tasks/orchestrator.py

@@ -153,48 +153,47 @@ def _run_request(
 
 def _set_request_cannot_be_run_status(
  request_type: ActivationRequest,
- process_parent_type: ProcessParentType,
- process_parent_id: int,
+ process_parent: Union[Activation, EventStream],
  status_text: str,
 ) -> None:
  """Set the request status to indicate it cannot be run on this node.
 
  The request cannot be run on this node due to detected functional issues.
- While it cannot be run on this node anothter functional node can run it.
- This method sets the request's state such that other nodes can identify
- it as something to be run, if possible. This identification is performed
- as part of monitor_rulebook_processes.
+ While it cannot be run on this node anothter functional node may be able
+ to.
  """
- status_manager = StatusManager(
- get_process_parent(process_parent_type, process_parent_id),
- )
-
  status = ActivationStatus.WORKERS_OFFLINE
  # If the request is start/auto-start we know the request isn't
- # running.
+ # running and can be put into the appropriate pending state.
+ # If a restart it can also be put in the appropriate pending
+ # state.
+ #
  # If the request is in one of the pending states that also
  # indicates it's not running; this check is for previous
  # failures to start that got dispatched from the monitor
  # task.
- if (
- request_type
- in [
-  ActivationRequest.START,
-  ActivationRequest.AUTO_START,
- ]
- ) or (
-  status_manager.db_instance.status
-  in [
-  ActivationStatus.PENDING,
-  ActivationStatus.PENDING_WORKERS_OFFLINE,
- ]
+ if request_type in [
+ ActivationRequest.START,
+ ActivationRequest.AUTO_START,
+ ActivationRequest.RESTART,
+ ]:
+ status = ActivationStatus.PENDING
+
+ # If the target status is WORKERS_OFFLINE and the process already has that
+ # status there's nothing to do. For any other situation we need to be
+ # certain to set the status text.
+ if (status != ActivationStatus.WORKERS_OFFLINE) or (
+ process_parent.status != ActivationStatus.WORKERS_OFFLINE
  ):
- status = ActivationStatus.PENDING_WORKERS_OFFLINE
+ status_manager = StatusManager(process_parent)
+ status_manager.set_status(status, status_text)
 
- status_manager.set_status(status, status_text)
- # There may not be a latest instance.
- if status_manager.latest_instance:
- status_manager.set_latest_instance_status(status, status_text)
+ # Set the latest instance's status only if there is a latest instance
+ # and only if the status is WORKERS_OFFLINE.
+ if status_manager.latest_instance and (
+ status == ActivationStatus.WORKERS_OFFLINE
+ ):
+ status_manager.set_latest_instance_status(status, status_text)
 
 
 def dispatch(
@@ -289,17 +288,13 @@ def dispatch(
  )
 
  if not enqueable:
- # The request is not enqueable. If not a restart set the request
- # status to "workers offline".
- if request_type != ActivationRequest.RESTART:
- # Set the request status such that, if possible, a functional
- # node can pick it up and run it.
- _set_request_cannot_be_run_status(
- request_type,
- process_parent_type,
- process_parent_id,
- status_msg,
- )
+ # The request is not enqueable. Set the request status such that, if
+ # possible, a functional node can pick it up and run it.
+ _set_request_cannot_be_run_status(
+ request_type,
+ get_process_parent(process_parent_type, process_parent_id),
+ status_msg,
+ )
  else:
  unique_enqueue(
  queue_name,
@@ -480,7 +475,6 @@ def monitor_rulebook_processes() -> None:
  for process in models.RulebookProcess.objects.filter(
  status__in=[
  ActivationStatus.RUNNING,
- ActivationStatus.PENDING_WORKERS_OFFLINE,
  ActivationStatus.WORKERS_OFFLINE,
  ]
  ):

tests/integration/api/test_activation.py

@@ -497,7 +497,6 @@ def test_enable_activation(
  for state in [
  enums.ActivationStatus.COMPLETED,
  enums.ActivationStatus.PENDING,
- enums.ActivationStatus.PENDING_WORKERS_OFFLINE,
  enums.ActivationStatus.STOPPED,
  enums.ActivationStatus.FAILED,
  ]:

tests/integration/core/test_rulebook_process.py

@@ -85,7 +85,6 @@ def test_rulebook_process_save(init_data):
  enums.ActivationStatus.UNRESPONSIVE,
  enums.ActivationStatus.ERROR,
  enums.ActivationStatus.PENDING,
- enums.ActivationStatus.PENDING_WORKERS_OFFLINE,
  ]:
  instance.status = status
  instance.save(update_fields=["status"])

tools/docker/docker-compose-dev.yaml

@@ -22,7 +22,7 @@ x-environment: &common-env
  EDA_SECRET_KEY: ${EDA_SECRET_KEY:-'insecure'}
  EDA_DEBUG: ${EDA_DEBUG:-True}
  EDA_DB_PASSWORD: ${EDA_DB_PASSWORD:-'secret'}
- EDA_RULEBOOK_WORKER_QUEUES: "activation-node1,activation-node2"
+ EDA_RULEBOOK_WORKER_QUEUES: "jbs-node1,jbs-node2"
  EDA_ANSIBLE_BASE_JWT_VALIDATE_CERT: ${EDA_ANSIBLE_BASE_JWT_VALIDATE_CERT:-False}
  EDA_ANSIBLE_BASE_JWT_KEY: ${EDA_ANSIBLE_BASE_JWT_KEY:-'https://localhost'}
 
@@ -197,7 +197,7 @@ services:
  image: "${EDA_IMAGE:-localhost/aap-eda}"
  environment:
  <<: *common-env
- EDA_RULEBOOK_QUEUE_NAME: 'activation-node2'
+ EDA_RULEBOOK_QUEUE_NAME: 'jbs-node2'
  EDA_PODMAN_SOCKET_URL: tcp://podman-node2:8888
  command:
  - aap-eda-manage