feat(core): dispatch durable dataflow runs

This commit is contained in:
2026-07-30 02:32:49 +02:00
parent ca8a8c5111
commit cf7afe9dda
4 changed files with 233 additions and 0 deletions

View File

@@ -7,7 +7,9 @@ from celery import Celery
from govoplan_core.core.campaigns import CAPABILITY_CAMPAIGNS_DELIVERY_TASKS, CampaignDeliveryTaskProvider
from govoplan_core.core.calendar import CAPABILITY_CALENDAR_OUTBOX, CalendarOutboxProvider
from govoplan_core.core.dataflows import (
CAPABILITY_DATAFLOW_RUN_WORKER,
CAPABILITY_DATAFLOW_TRIGGER_DISPATCHER,
DataflowRunWorker,
DataflowTriggerDispatcher,
)
from govoplan_core.core.events import (
@@ -42,6 +44,8 @@ celery.conf.update(
"govoplan.notifications.deliver": {"queue": "notifications"},
"govoplan.notifications.deliver_pending": {"queue": "notifications"},
"govoplan.calendar.dispatch_outbox": {"queue": "calendar"},
"govoplan.dataflow.dispatch_runs": {"queue": "dataflow"},
"govoplan.dataflow.purge_runs": {"queue": "dataflow"},
"govoplan.dataflow.dispatch_triggers": {"queue": "dataflow"},
"govoplan.events.dispatch_outbox": {"queue": "events"},
"govoplan.events.purge_outbox": {"queue": "events"},
@@ -60,6 +64,16 @@ celery.conf.update(
"schedule": 60.0,
"args": (100,),
},
"dataflow-runs-every-five-seconds": {
"task": "govoplan.dataflow.dispatch_runs",
"schedule": 5.0,
"args": (10,),
},
"dataflow-run-retention-daily": {
"task": "govoplan.dataflow.purge_runs",
"schedule": 24 * 60 * 60.0,
"args": (500,),
},
"platform-events-every-ten-seconds": {
"task": "govoplan.events.dispatch_outbox",
"schedule": 10.0,
@@ -131,6 +145,18 @@ def _dataflow_trigger_dispatcher(
return capability
def _dataflow_run_worker(
registry: PlatformRegistry | None = None,
) -> DataflowRunWorker | None:
registry = registry or _platform_registry()
if not registry.has_capability(CAPABILITY_DATAFLOW_RUN_WORKER):
return None
capability = registry.require_capability(CAPABILITY_DATAFLOW_RUN_WORKER)
if not isinstance(capability, DataflowRunWorker):
raise RuntimeError("Dataflow run worker capability is invalid")
return capability
def _platform_event_outbox(
registry: PlatformRegistry | None = None,
) -> PlatformEventOutbox | None:
@@ -234,6 +260,56 @@ def dispatch_dataflow_triggers(self, limit: int = 100):
return result
@celery.task(
name="govoplan.dataflow.dispatch_runs",
bind=True,
max_retries=0,
)
def dispatch_dataflow_runs(self, limit: int = 10):
"""Claim and execute durable Dataflow runs outside the API process."""
from govoplan_core.db.session import get_database
with get_database().SessionLocal() as session:
provider = _dataflow_run_worker()
if provider is None:
return {
"claimed": 0,
"succeeded": 0,
"retrying": 0,
"failed": 0,
"cancelled": 0,
}
result = dict(
provider.dispatch_pending(
session,
limit=limit,
worker_id=getattr(self.request, "hostname", None),
)
)
session.commit()
return result
@celery.task(
name="govoplan.dataflow.purge_runs",
bind=True,
max_retries=0,
)
def purge_dataflow_runs(self, limit: int = 500):
"""Apply Dataflow evidence-retention policy without deleting run records."""
from govoplan_core.db.session import get_database
with get_database().SessionLocal() as session:
provider = _dataflow_run_worker()
if provider is None:
return {"purged": 0}
result = dict(provider.purge_expired(session, limit=limit))
session.commit()
return result
@celery.task(
name="govoplan.events.dispatch_outbox",
bind=True,