gustavosantos/GTI_GERENCIADOR_FISCAL
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: zero-touch Agent — central scheduler, central pause, delta sync, discovery-driven installer

Browse Source 
Hub: installations.processing_enabled / sync_interval_seconds / last_sync_* (migration 0026).
The first heartbeat of an installation without coverage queues exactly one INITIAL_SYNC;
afterwards one SYNC_INCREMENTAL per interval (idempotent, backs off on failure). One fiscal
job at a time per Agent (FIFO, stale RUNNING recovery). Job completion now updates the
coverage snapshot, blockers and JOB_STARTED/COMPLETED/FAILED events. Panel: Sincronizar
agora / Reprocessar tudo / Pausar / Retomar + status; pause is enforced Hub-side.
Agent: local AGENT_JOBS_ENABLED is ignored (central control); SYNC_INCREMENTAL re-classifies
only new/changed products via a per-product fingerprint+decision cache and rebuilds coverage
from it; results carry operational summary and blockers; Firebird credentials come from a
DPAPI firebird.bin. Setup: read-only SGV Firebird discovery (auto-select only when
unambiguous, wizard chooser otherwise), starts the service, waits for the first heartbeat,
upgrade-in-place mode. All jobs stay read-only (FIREBIRD_WRITES=0). Version 1.1.3.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This commit is contained in:
gti
2026-09-24 08:47:00 -03:00
co-authored by Claude Sonnet 5
parent 9f9e06e9e9
commit bb663a979c
36 changed files with 1409 additions and 118 deletions
Download Patch File Download Diff File Expand all files Collapse all files
central/app/api.py
+46 -2
View File
@@ -10,7 +10,7 @@ from .auth import authenticate, bootstrap
from .classification import ENGINE_VERSION, Classifier
from .config import settings
from .datasets.service import DatasetService
from .jobs import claim_next_job, complete_job, create_job, fail_job
from .jobs import SYNC_KINDS, claim_next_job, complete_job, create_job, fail_job, schedule_automatic_sync
from .blocked_products import list_blockers, resolve_blocker, resolve_product_blockers, upsert_blocker
from .context_keys import validate_entry
from .models import AgentSnapshot, BlockedProduct, FiscalContextEntry, ClassificationCache, ClassificationRequest, ClassificationResult, Dataset, Installation, Job
@@ -206,8 +206,13 @@ def router(session_factory, bootstrap_token):
session.add(agent); session.flush()
record_event(session, tenant_id=agent.tenant_id, installation_id=installation.id, agent_id=agent.id, severity="INFO", event_type="HEARTBEAT", message="Agent heartbeat", context={"status": agent.last_status, "agent_version": agent.agent_version})
session.commit()
# Central scheduler: exactly one INITIAL_SYNC after the first heartbeat, then
# one SYNC_INCREMENTAL per interval — idempotent, no local cron/toggle needed.
if agent.enabled:
schedule_automatic_sync(session, agent, installation, snapshot)
logger.info("heartbeat agent_id=%s status=%s last_job_id=%s", agent.id, agent.last_status, payload.get("last_job_id"))
return {"agent_id": agent.id, "installation_id": installation.installation_id, "last_seen_at": agent.last_seen_at.isoformat(), "enabled": agent.enabled}
return {"agent_id": agent.id, "installation_id": installation.installation_id, "last_seen_at": agent.last_seen_at.isoformat(), "enabled": agent.enabled,
"processing_enabled": bool(installation.processing_enabled), "sync_interval_seconds": installation.sync_interval_seconds}
finally: session.close()
@routes.post("/agents/register")
@@ -332,6 +337,36 @@ def router(session_factory, bootstrap_token):
"attempts": job.attempts, "result_summary": job.result_summary, "error": job.error,
}
def _apply_job_outcome(session, agent, installation, job, result_summary, *, failed):
"""After a job finishes: coverage snapshot (from the Agent's operational
summary), sync bookkeeping and an audit-friendly event. Never stores secrets."""
now = datetime.now(timezone.utc)
snapshot = session.scalar(select(AgentSnapshot).where(AgentSnapshot.installation_id == installation.id))
if snapshot is None:
snapshot = AgentSnapshot(installation_id=installation.id); session.add(snapshot)
snapshot.last_job_status = "FAILED" if failed else "COMPLETED"
snapshot.last_job_at = now
context = {"kind": job.kind}
operational = result_summary.get("operational") if not failed else None
if operational and job.kind in SYNC_KINDS:
total = int(operational.get("TOTAL_PRODUCTS", 0))
complete = int(operational.get("COMPLETE", 0)) + int(operational.get("CORRECTED", 0))
snapshot.total_products = total
snapshot.fiscally_complete = complete
snapshot.safe_blocked = int(operational.get("SAFE_BLOCKED", 0))
snapshot.failed_products = int(operational.get("FAILED_TECHNICAL", 0))
snapshot.coverage_percent = round(100.0 * complete / total, 2) if total else 0.0
snapshot.blockers = {state: int(count) for state, count in operational.items() if state.startswith("BLOCKED_") and count}
snapshot.last_successful_job_at = now
installation.last_sync_at = now
installation.last_sync_changed = int(result_summary.get("products_changed", result_summary.get("products_processed", 0)) or 0)
context.update({"total_products": total, "products_changed": installation.last_sync_changed, "products_processed": result_summary.get("products_processed")})
if failed:
context["error"] = result_summary.get("error")
record_event(session, tenant_id=agent.tenant_id, installation_id=installation.id, agent_id=agent.id, severity="ERROR" if failed else "INFO",
event_type="JOB_FAILED" if failed else "JOB_COMPLETED", message=f"Job {job.kind} {'failed' if failed else 'completed'}", context=context, job_id=job.id)
session.commit()
@routes.post("/jobs")
def create_job_route(payload: dict, identity=Depends(principal)):
session, agent = identity
@@ -353,6 +388,10 @@ def router(session_factory, bootstrap_token):
try:
job = claim_next_job(session, agent)
if job is None: return None
installation = session.scalar(select(Installation).where(Installation.agent_id == agent.id))
if installation is not None:
record_event(session, tenant_id=agent.tenant_id, installation_id=installation.id, agent_id=agent.id, severity="INFO", event_type="JOB_STARTED", message=f"Job {job.kind} started", context={"kind": job.kind}, job_id=job.id)
session.commit()
logger.info("job claimed job_id=%s agent_id=%s kind=%s", job.id, agent.id, job.kind)
return _job_view(job)
finally: session.close()
@@ -388,6 +427,8 @@ def router(session_factory, bootstrap_token):
for product_code in result_summary.get("resolved_products", []):
resolve_product_blockers(session, agent.tenant_id, product_code, installation_id=installation.id if installation else None)
if not complete_job(session, job, result_summary): raise HTTPException(409, "job is not RUNNING")
if installation is not None:
_apply_job_outcome(session, agent, installation, job, result_summary, failed=False)
logger.info("job completed job_id=%s agent_id=%s result=%s", job_id, agent.id, payload.get("result_summary", {}))
return _job_view(session.get(Job, job_id))
finally: session.close()
@@ -399,6 +440,9 @@ def router(session_factory, bootstrap_token):
job = session.get(Job, job_id)
if job is None or job.agent_id != agent.id: raise HTTPException(404, "job not found")
if not fail_job(session, job, payload.get("error", "unknown error")): raise HTTPException(409, "job is not RUNNING")
installation = session.scalar(select(Installation).where(Installation.agent_id == agent.id))
if installation is not None:
_apply_job_outcome(session, agent, installation, job, {"error": str(payload.get("error", "unknown error"))[:500]}, failed=True)
logger.warning("job failed job_id=%s agent_id=%s error=%s", job_id, agent.id, payload.get("error", "unknown error"))
return _job_view(session.get(Job, job_id))
finally: session.close()