feat(azure-ai+gpu-ui): per-tenant Azure AI config + GPU health panel

- Per-tenant Azure AI config stored in tenants.tenant_config JSONB
- GET/PUT /api/tenants/{id}/ai-config + POST .../test connection
- api_key never returned to frontend (has_api_key: bool pattern)
- azure_ai.py resolves creds from tenant config when ai_enabled=True
- ai_tasks.py loads tenant config and passes it to validate_thumbnail
- Admin GPU Status section: probe button + status badge + last-checked time
- Notifications: _BELL_CHANNELS filter (notification+alert only in bell)
- Tenants.tsx: per-row Azure AI Config modal with URL auto-parse helper
- Remove duplicate in-memory /gpu-probe endpoints (kept DB-backed /probe/gpu)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
2026-03-08 21:04:09 +01:00
parent 34f89cc225
commit 22c29d5655
11 changed files with 792 additions and 24 deletions
+21 -6
View File
@@ -51,30 +51,42 @@ def _visibility_filter(user: User):
return and_(AuditLog.notification == True, targeted) # noqa: E712
# Default channels shown in bell dropdown
_BELL_CHANNELS = ("notification", "alert")
@router.get("", response_model=NotificationListResponse)
async def list_notifications(
limit: int = Query(20, ge=1, le=100),
offset: int = Query(0, ge=0),
unread_only: bool = Query(False),
channel: Optional[str] = Query(None, description="Filter by channel: notification, activity, alert. Defaults to notification+alert."),
user: User = Depends(get_current_user),
db: AsyncSession = Depends(get_db),
):
vis = _visibility_filter(user)
# Channel filter
if channel:
channel_filter = AuditLog.channel == channel
else:
channel_filter = AuditLog.channel.in_(_BELL_CHANNELS)
# Total count
total_q = select(func.count(AuditLog.id)).where(vis)
total_q = select(func.count(AuditLog.id)).where(vis, channel_filter)
if unread_only:
total_q = total_q.where(AuditLog.read_at.is_(None))
total = (await db.execute(total_q)).scalar() or 0
# Unread count (always)
unread_q = select(func.count(AuditLog.id)).where(vis, AuditLog.read_at.is_(None))
# Unread count (always — only for bell channels, not activity)
bell_channel_filter = AuditLog.channel.in_(_BELL_CHANNELS)
unread_q = select(func.count(AuditLog.id)).where(vis, bell_channel_filter, AuditLog.read_at.is_(None))
unread_count = (await db.execute(unread_q)).scalar() or 0
# Items
items_q = (
select(AuditLog)
.where(vis)
.where(vis, channel_filter)
.order_by(AuditLog.timestamp.desc())
.offset(offset)
.limit(limit)
@@ -105,7 +117,8 @@ async def unread_count(
db: AsyncSession = Depends(get_db),
):
vis = _visibility_filter(user)
q = select(func.count(AuditLog.id)).where(vis, AuditLog.read_at.is_(None))
bell_channel_filter = AuditLog.channel.in_(_BELL_CHANNELS)
q = select(func.count(AuditLog.id)).where(vis, bell_channel_filter, AuditLog.read_at.is_(None))
count = (await db.execute(q)).scalar() or 0
return UnreadCountResponse(unread_count=count)
@@ -182,7 +195,9 @@ async def update_my_notification_config(
):
if channel not in ("in_app", "email"):
raise HTTPException(status_code=400, detail="channel must be 'in_app' or 'email'")
return await upsert_notification_config(db, current_user.id, event_type, channel, body.enabled)
if body.frequency is not None and body.frequency not in ("immediate", "daily", "never"):
raise HTTPException(status_code=400, detail="frequency must be 'immediate', 'daily', or 'never'")
return await upsert_notification_config(db, current_user.id, event_type, channel, body.enabled, body.frequency)
@router.post("/config/reset", response_model=list[NotificationConfigOut])
+1
View File
@@ -683,3 +683,4 @@ async def update_worker_config(
enabled=cfg.enabled,
updated_at=cfg.updated_at.isoformat(),
)
+9
View File
@@ -12,6 +12,15 @@ _DEFAULT_TENANT_CONFIG = {
"fallback_material": "SCHAEFFLER_059999_FailedMaterial",
"notifications_enabled": True,
"invoice_prefix": "INV",
# Azure AI validation (per-tenant)
"ai_enabled": False,
"ai_api_key": None, # stored but never returned to frontend
"ai_endpoint": None, # e.g. https://myinstance.openai.azure.com
"ai_deployment": "gpt-4o",
"ai_api_version": "2024-02-01",
"ai_max_tokens": 500,
"ai_temperature": 0.1,
"ai_validation_prompt": None, # None = use DEFAULT_VALIDATION_PROMPT from azure_ai.py
}
+124 -1
View File
@@ -1,11 +1,16 @@
import uuid
from fastapi import APIRouter, Depends, HTTPException, status
from sqlalchemy.ext.asyncio import AsyncSession
from sqlalchemy import update
from app.database import get_db
from app.utils.auth import require_admin
from app.domains.tenants.schemas import TenantCreate, TenantUpdate, TenantOut
from app.domains.tenants.schemas import (
TenantCreate, TenantUpdate, TenantOut,
TenantAIConfigUpdate, TenantAIConfigOut,
)
from app.domains.tenants import service
from app.domains.tenants.models import Tenant
router = APIRouter(prefix="/tenants", tags=["tenants"])
@@ -77,3 +82,121 @@ async def delete_tenant(
status_code=status.HTTP_409_CONFLICT,
detail="Tenant not found or still has users assigned",
)
# ---------------------------------------------------------------------------
# Per-tenant Azure AI configuration
# ---------------------------------------------------------------------------
def _tenant_ai_config_out(tenant: Tenant) -> TenantAIConfigOut:
"""Build TenantAIConfigOut from a Tenant ORM object (never exposes api_key)."""
cfg = tenant.tenant_config or {}
return TenantAIConfigOut(
ai_enabled=cfg.get("ai_enabled", False),
ai_endpoint=cfg.get("ai_endpoint"),
ai_deployment=cfg.get("ai_deployment", "gpt-4o"),
ai_api_version=cfg.get("ai_api_version", "2024-02-01"),
has_api_key=bool(cfg.get("ai_api_key")),
ai_max_tokens=cfg.get("ai_max_tokens", 500),
ai_temperature=cfg.get("ai_temperature", 0.1),
ai_validation_prompt=cfg.get("ai_validation_prompt"),
)
@router.get("/{tenant_id}/ai-config", response_model=TenantAIConfigOut)
async def get_tenant_ai_config(
tenant_id: uuid.UUID,
db: AsyncSession = Depends(get_db),
_: object = Depends(require_admin),
):
"""Return AI config for a tenant (without the raw api_key)."""
tenant = await service.get_tenant(db, tenant_id)
if not tenant:
raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Tenant not found")
return _tenant_ai_config_out(tenant)
@router.put("/{tenant_id}/ai-config", response_model=TenantAIConfigOut)
async def update_tenant_ai_config(
tenant_id: uuid.UUID,
body: TenantAIConfigUpdate,
db: AsyncSession = Depends(get_db),
_: object = Depends(require_admin),
):
"""Merge AI configuration into tenant_config JSONB.
If ai_api_key is None in the request body, the existing key is preserved.
"""
tenant = await service.get_tenant(db, tenant_id)
if not tenant:
raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Tenant not found")
# Build the merged config dict
current_cfg: dict = dict(tenant.tenant_config or {})
current_cfg["ai_enabled"] = body.ai_enabled
current_cfg["ai_endpoint"] = body.ai_endpoint
current_cfg["ai_deployment"] = body.ai_deployment
current_cfg["ai_api_version"] = body.ai_api_version
current_cfg["ai_max_tokens"] = body.ai_max_tokens
current_cfg["ai_temperature"] = body.ai_temperature
current_cfg["ai_validation_prompt"] = body.ai_validation_prompt
# Only overwrite the stored key if a new value was actually provided
if body.ai_api_key is not None:
current_cfg["ai_api_key"] = body.ai_api_key
# Use direct SQL UPDATE to reliably persist JSONB mutations
await db.execute(
update(Tenant)
.where(Tenant.id == tenant_id)
.values(tenant_config=current_cfg)
)
await db.commit()
await db.refresh(tenant)
return _tenant_ai_config_out(tenant)
@router.post("/{tenant_id}/ai-config/test")
async def test_tenant_ai_config(
tenant_id: uuid.UUID,
db: AsyncSession = Depends(get_db),
_: object = Depends(require_admin),
):
"""Send a minimal ping to Azure OpenAI using the tenant's stored credentials.
Returns {"ok": true} or {"ok": false, "error": "human readable message"}.
"""
tenant = await service.get_tenant(db, tenant_id)
if not tenant:
raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Tenant not found")
cfg = tenant.tenant_config or {}
api_key = cfg.get("ai_api_key")
endpoint = cfg.get("ai_endpoint")
deployment = cfg.get("ai_deployment", "gpt-4o")
api_version = cfg.get("ai_api_version", "2024-02-01")
if not api_key:
return {"ok": False, "error": "No API key configured for this tenant"}
if not endpoint:
return {"ok": False, "error": "No endpoint configured for this tenant"}
import asyncio
def _ping() -> dict:
try:
from openai import AzureOpenAI
client = AzureOpenAI(
api_key=api_key,
azure_endpoint=endpoint,
api_version=api_version,
)
client.chat.completions.create(
model=deployment,
messages=[{"role": "user", "content": "ping"}],
max_tokens=5,
)
return {"ok": True}
except Exception as exc:
return {"ok": False, "error": str(exc)}
return await asyncio.to_thread(_ping)
+22
View File
@@ -24,3 +24,25 @@ class TenantOut(BaseModel):
created_at: datetime
model_config = {"from_attributes": True}
class TenantAIConfigUpdate(BaseModel):
ai_enabled: bool = False
ai_endpoint: str | None = None
ai_deployment: str = "gpt-4o"
ai_api_version: str = "2024-02-01"
ai_api_key: str | None = None # optional — don't require re-entry
ai_max_tokens: int = 500
ai_temperature: float = 0.1
ai_validation_prompt: str | None = None
class TenantAIConfigOut(BaseModel):
ai_enabled: bool
ai_endpoint: str | None
ai_deployment: str
ai_api_version: str
has_api_key: bool # True if ai_api_key is set — never return the key itself
ai_max_tokens: int
ai_temperature: float
ai_validation_prompt: str | None
+42 -12
View File
@@ -24,10 +24,13 @@ Respond in JSON with exactly these fields:
}"""
def validate_thumbnail(order_item_id: str) -> dict:
def validate_thumbnail(order_item_id: str, tenant_config: dict | None = None) -> dict:
"""
Validate thumbnail orientation using Azure GPT-4o Vision.
Updates the order_item AI validation fields in DB.
If tenant_config is provided and tenant_config["ai_enabled"] is True,
the tenant's own Azure credentials are used instead of global settings.
"""
from app.config import settings
from sqlalchemy import create_engine
@@ -45,7 +48,7 @@ def validate_thumbnail(order_item_id: str) -> dict:
session.commit()
try:
result = _call_azure_vision(item.thumbnail_path, settings)
result = _call_azure_vision(item.thumbnail_path, settings, tenant_config)
item.ai_validation_status = AIValidationStatus.completed
item.ai_validation_result = result
except Exception as exc:
@@ -58,11 +61,38 @@ def validate_thumbnail(order_item_id: str) -> dict:
return result
def _call_azure_vision(thumbnail_path: str | None, settings) -> dict:
"""Call Azure OpenAI GPT-4o with a base64-encoded thumbnail."""
def _call_azure_vision(
thumbnail_path: str | None,
settings,
tenant_config: dict | None = None,
) -> dict:
"""Call Azure OpenAI GPT-4o with a base64-encoded thumbnail.
Credential resolution order:
1. tenant_config (if provided and ai_enabled=True)
2. Global settings (azure_openai_* env vars)
"""
import json
if not settings.azure_openai_api_key or not settings.azure_openai_endpoint:
# Resolve credentials from tenant config or global settings
if tenant_config and tenant_config.get("ai_enabled"):
api_key = tenant_config.get("ai_api_key") or settings.azure_openai_api_key
endpoint = tenant_config.get("ai_endpoint") or settings.azure_openai_endpoint
deployment = tenant_config.get("ai_deployment") or settings.azure_openai_deployment
api_version = tenant_config.get("ai_api_version") or settings.azure_openai_api_version
max_tokens = int(tenant_config.get("ai_max_tokens", 500))
temperature = float(tenant_config.get("ai_temperature", 0.1))
prompt = tenant_config.get("ai_validation_prompt") or VALIDATION_PROMPT
else:
api_key = settings.azure_openai_api_key
endpoint = settings.azure_openai_endpoint
deployment = settings.azure_openai_deployment
api_version = settings.azure_openai_api_version
max_tokens = 500
temperature = 0.1
prompt = VALIDATION_PROMPT
if not api_key or not endpoint:
raise ValueError("Azure OpenAI credentials not configured")
if not thumbnail_path or not Path(thumbnail_path).exists():
@@ -72,21 +102,21 @@ def _call_azure_vision(thumbnail_path: str | None, settings) -> dict:
from openai import AzureOpenAI
client = AzureOpenAI(
api_key=settings.azure_openai_api_key,
azure_endpoint=settings.azure_openai_endpoint,
api_version=settings.azure_openai_api_version,
api_key=api_key,
azure_endpoint=endpoint,
api_version=api_version,
)
with open(thumbnail_path, "rb") as f:
image_b64 = base64.b64encode(f.read()).decode("utf-8")
response = client.chat.completions.create(
model=settings.azure_openai_deployment,
model=deployment,
messages=[
{
"role": "user",
"content": [
{"type": "text", "text": VALIDATION_PROMPT},
{"type": "text", "text": prompt},
{
"type": "image_url",
"image_url": {"url": f"data:image/png;base64,{image_b64}"},
@@ -94,8 +124,8 @@ def _call_azure_vision(thumbnail_path: str | None, settings) -> dict:
],
}
],
max_tokens=500,
temperature=0.1,
max_tokens=max_tokens,
temperature=temperature,
)
content = response.choices[0].message.content or ""
+25 -2
View File
@@ -7,11 +7,34 @@ logger = logging.getLogger(__name__)
@celery_app.task(bind=True, name="app.tasks.ai_tasks.validate_item", queue="ai_validation")
def validate_item(self, order_item_id: str):
"""Validate orientation of a rendered thumbnail via Azure GPT-4o Vision."""
"""Validate orientation of a rendered thumbnail via Azure GPT-4o Vision.
Loads the order item's tenant config and passes it to validate_thumbnail()
so that per-tenant Azure credentials are used when configured.
"""
logger.info(f"AI validation for item: {order_item_id}")
try:
from app.config import settings
from sqlalchemy import create_engine
from sqlalchemy.orm import Session
from app.models.order_item import OrderItem
from app.services.azure_ai import validate_thumbnail
validate_thumbnail(order_item_id)
# Load tenant config for this order item
tenant_config: dict | None = None
try:
engine = create_engine(settings.database_url_sync)
with Session(engine) as session:
item = session.get(OrderItem, __import__("uuid").UUID(order_item_id))
if item and hasattr(item, "order") and item.order and item.order.tenant_id:
from app.domains.tenants.models import Tenant
tenant = session.get(Tenant, item.order.tenant_id)
if tenant:
tenant_config = tenant.tenant_config or {}
except Exception as exc:
logger.warning(f"Could not load tenant config for {order_item_id}: {exc}")
validate_thumbnail(order_item_id, tenant_config=tenant_config)
except Exception as exc:
logger.error(f"AI validation failed for {order_item_id}: {exc}")
raise self.retry(exc=exc, countdown=30, max_retries=3)