Fix purge 500 when SQLite database is locked (Spec 017).

Enable WAL/busy_timeout, retry writes, reject duplicate running jobs with HTTP 409,
use bcrypt directly instead of broken passlib 1.7.4 + bcrypt 4.x, and improve UI errors.

projects/ops-desk/api/app/auth.py

@@ -13,7 +13,7 @@ from typing import Any
 
 from fastapi import Depends, Header, HTTPException, Request
 from jose import JWTError, jwt
-from passlib.context import CryptContext
+import bcrypt
 
 from app.totp_util import verify_code as verify_totp_code
 
@@ -26,8 +26,6 @@ DESK_BOOTSTRAP_PASSWORD = os.getenv("DESK_BOOTSTRAP_PASSWORD", "805353")
 AUTH_LOGIN_RATE_LIMIT = int(os.getenv("AUTH_LOGIN_RATE_LIMIT", "5"))
 OPS_INTERNAL_TOKEN = os.getenv("OPS_INTERNAL_TOKEN", "")
 
-pwd_context = CryptContext(schemes=["bcrypt"], deprecated="auto")
-
 _login_attempts: dict[str, list[float]] = {}
 _mfa_pending: dict[str, tuple[str, float]] = {}
 MFA_TOKEN_TTL_SEC = 300
@@ -54,17 +52,24 @@ class DeskUser:
 
 def db() -> sqlite3.Connection:
     DB_PATH.parent.mkdir(parents=True, exist_ok=True)
-    conn = sqlite3.connect(DB_PATH)
+    conn = sqlite3.connect(DB_PATH, timeout=30.0)
     conn.row_factory = sqlite3.Row
+    conn.execute("PRAGMA journal_mode=WAL")
+    conn.execute("PRAGMA busy_timeout=30000")
     return conn
 
 
 def hash_password(password: str) -> str:
-    return pwd_context.hash(password)
+    return bcrypt.hashpw(password.encode("utf-8"), bcrypt.gensalt()).decode("utf-8")
 
 
 def verify_password(plain: str, hashed: str) -> bool:
-    return pwd_context.verify(plain, hashed)
+    if not plain or not hashed:
+        return False
+    try:
+        return bcrypt.checkpw(plain.encode("utf-8"), hashed.encode("utf-8"))
+    except (ValueError, TypeError):
+        return False
 
 
 def init_auth_schema(conn: sqlite3.Connection) -> None:

projects/ops-desk/api/app/main.py

@@ -133,8 +133,10 @@ TICKET_COLUMNS = "id,tenant_id,subject,status,payload,created_at,assigned_to,ass
 
 def db():
     DB_PATH.parent.mkdir(parents=True, exist_ok=True)
-    conn = sqlite3.connect(DB_PATH)
+    conn = sqlite3.connect(DB_PATH, timeout=30.0)
     conn.row_factory = sqlite3.Row
+    conn.execute("PRAGMA journal_mode=WAL")
+    conn.execute("PRAGMA busy_timeout=30000")
     return conn
 
 
@@ -172,6 +174,11 @@ def init_db():
         from app import billing_store
         migration_store.init_schema(conn)
         billing_store.init_schema(conn)
+        from app.vm112_purge_jobs import init_purge_jobs_schema
+
+        init_purge_jobs_schema(conn)
+        conn.execute("PRAGMA journal_mode=WAL")
+        conn.execute("PRAGMA busy_timeout=30000")
         conn.commit()
 
 

projects/ops-desk/api/app/vm112_domains_routes.py

@@ -2,6 +2,8 @@
 
 from __future__ import annotations
 
+import sqlite3
+
 from fastapi import APIRouter, Depends, HTTPException
 from fastapi.responses import StreamingResponse
 from pydantic import BaseModel, Field
@@ -9,7 +11,13 @@ from pydantic import BaseModel, Field
 from app import auth, vm112_domains
 from app.permissions import can_manage_vm112_domains
 from app.vm112_purge_stream import purge_sse_generator
-from app.vm112_purge_jobs import get_job_public, list_jobs, recover_job, start_job
+from app.vm112_purge_jobs import (
+    PurgeJobConflictError,
+    get_job_public,
+    list_jobs,
+    recover_job,
+    start_job,
+)
 
 router = APIRouter(prefix="/api/v1/vm112", tags=["vm112-domains"])
 
@@ -116,7 +124,17 @@ def start_purge_job(
 ):
     """Inicia purge em background; consultar GET /purge/jobs/{id} (recomendado via Traefik)."""
     domain = _validate_purge_request(domain, body)
+    try:
         job_id = start_job(domain, body.root_password, user.username)
+    except PurgeJobConflictError as exc:
+        raise HTTPException(409, str(exc)) from exc
+    except sqlite3.OperationalError as exc:
+        if "locked" in str(exc).lower():
+            raise HTTPException(
+                503,
+                "Base de dados ocupada — aguarde o purge em curso ou tente novamente em alguns segundos",
+            ) from exc
+        raise HTTPException(500, f"Erro SQLite ao iniciar purge: {exc}") from exc
     return {"ok": True, "job_id": job_id, "domain": domain, "status": "running"}
 
 

projects/ops-desk/api/app/vm112_purge_jobs.py

@@ -3,7 +3,9 @@
 from __future__ import annotations
 
 import json
+import sqlite3
 import threading
+import time
 import traceback
 import uuid
 from datetime import datetime, timezone
@@ -12,6 +14,11 @@ from typing import Any, Callable
 from app import auth, vm112_domains
 
 _lock = threading.Lock()
+_schema_ready = False
+
+
+class PurgeJobConflictError(Exception):
+    """Já existe purge queued/running para o domínio."""
 
 
 def _now() -> str:
@@ -36,17 +43,42 @@ def init_purge_jobs_schema(conn) -> None:
         )
         """
     )
+    conn.execute(
+        """
+        CREATE INDEX IF NOT EXISTS idx_vm112_purge_jobs_domain_status
+        ON vm112_purge_jobs(domain, status)
+        """
+    )
     conn.commit()
 
 
 def _ensure_schema() -> None:
+    global _schema_ready
+    if _schema_ready:
+        return
     conn = auth.db()
     try:
         init_purge_jobs_schema(conn)
+        _schema_ready = True
     finally:
         conn.close()
 
 
+def _sqlite_retry(fn: Callable[[], Any], *, attempts: int = 8) -> Any:
+    last_err: Exception | None = None
+    for attempt in range(attempts):
+        try:
+            return fn()
+        except sqlite3.OperationalError as exc:
+            last_err = exc
+            if "locked" not in str(exc).lower() or attempt >= attempts - 1:
+                raise
+            time.sleep(min(0.05 * (2**attempt), 1.0))
+    if last_err:
+        raise last_err
+    return None
+
+
 def _row_to_job(row) -> dict[str, Any]:
     return {
         "id": row["id"],
@@ -66,6 +98,8 @@ def _row_to_job(row) -> dict[str, Any]:
 
 def _persist_job(job: dict[str, Any]) -> None:
     _ensure_schema()
+
+    def _write() -> None:
         conn = auth.db()
         try:
             job["updated_at"] = _now()
@@ -103,9 +137,13 @@ def _persist_job(job: dict[str, Any]) -> None:
         finally:
             conn.close()
 
+    _sqlite_retry(_write)
+
 
 def _load_job(job_id: str) -> dict[str, Any] | None:
     _ensure_schema()
+
+    def _read() -> dict[str, Any] | None:
         conn = auth.db()
         try:
             row = conn.execute(
@@ -115,15 +153,69 @@ def _load_job(job_id: str) -> dict[str, Any] | None:
         finally:
             conn.close()
 
+    return _sqlite_retry(_read)
+
+
+def _find_active_job_locked(conn: sqlite3.Connection, domain: str) -> sqlite3.Row | None:
+    return conn.execute(
+        """
+        SELECT id, domain, status, created_at FROM vm112_purge_jobs
+        WHERE domain = ? AND status IN ('queued', 'running')
+        ORDER BY created_at DESC LIMIT 1
+        """,
+        (domain.lower().strip(),),
+    ).fetchone()
+
 
 def _mutate_job(job_id: str, fn: Callable[[dict[str, Any]], None]) -> dict[str, Any] | None:
     with _lock:
-        job = _load_job(job_id)
+        def _mutate() -> dict[str, Any] | None:
-        if not job:
+            conn = auth.db()
+            try:
+                row = conn.execute(
+                    "SELECT * FROM vm112_purge_jobs WHERE id = ?", (job_id,)
+                ).fetchone()
+                if not row:
                     return None
+                job = _row_to_job(row)
                 fn(job)
-        _persist_job(job)
+                job["updated_at"] = _now()
+                conn.execute(
+                    """
+                    INSERT INTO vm112_purge_jobs (
+                        id, domain, status, timeline_json, elapsed_vm112,
+                        desk_json, vm112_json, error, by_user, created_at, updated_at
+                    ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+                    ON CONFLICT(id) DO UPDATE SET
+                        status = excluded.status,
+                        timeline_json = excluded.timeline_json,
+                        elapsed_vm112 = excluded.elapsed_vm112,
+                        desk_json = excluded.desk_json,
+                        vm112_json = excluded.vm112_json,
+                        error = excluded.error,
+                        by_user = excluded.by_user,
+                        updated_at = excluded.updated_at
+                    """,
+                    (
+                        job["id"],
+                        job["domain"],
+                        job["status"],
+                        json.dumps(job.get("timeline") or [], ensure_ascii=False),
+                        int(job.get("elapsed_vm112") or 0),
+                        json.dumps(job.get("desk") or {}, ensure_ascii=False),
+                        json.dumps(job.get("vm112") or {}, ensure_ascii=False),
+                        job.get("error"),
+                        job.get("by"),
+                        job.get("created_at") or _now(),
+                        job["updated_at"],
+                    ),
+                )
+                conn.commit()
                 return dict(job)
+            finally:
+                conn.close()
+
+        return _sqlite_retry(_mutate)
 
 
 def _upsert_step(job_id: str, step: dict[str, str]) -> None:
@@ -143,12 +235,13 @@ def _set_job(job_id: str, **fields: Any) -> None:
 
 
 def create_job(domain: str, username: str) -> str:
+    domain = domain.lower().strip()
     job_id = uuid.uuid4().hex[:16]
     now = _now()
     job = {
         "id": job_id,
         "job_id": job_id,
-        "domain": domain.lower().strip(),
+        "domain": domain,
         "status": "queued",
         "timeline": [],
         "elapsed_vm112": 0,
@@ -159,8 +252,43 @@ def create_job(domain: str, username: str) -> str:
         "created_at": now,
         "updated_at": now,
     }
+
+    def _create() -> None:
+        conn = auth.db()
+        try:
+            active = _find_active_job_locked(conn, domain)
+            if active:
+                raise PurgeJobConflictError(
+                    f"Purge já em curso para {domain} (job {active['id']}, status {active['status']})"
+                )
+            conn.execute(
+                """
+                INSERT INTO vm112_purge_jobs (
+                    id, domain, status, timeline_json, elapsed_vm112,
+                    desk_json, vm112_json, error, by_user, created_at, updated_at
+                ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+                """,
+                (
+                    job["id"],
+                    job["domain"],
+                    job["status"],
+                    "[]",
+                    0,
+                    "{}",
+                    "{}",
+                    None,
+                    job["by"],
+                    job["created_at"],
+                    job["updated_at"],
+                ),
+            )
+            conn.commit()
+        finally:
+            conn.close()
+
     with _lock:
-        _persist_job(job)
+        _ensure_schema()
+        _sqlite_retry(_create)
     return job_id
 
 

projects/ops-desk/frontend/assets/servicos.js

@@ -140,6 +140,16 @@ const DeskServices = (() => {
       let errText = typeof detail === 'string' ? detail : JSON.stringify(detail || `${res.status}`);
       if (res.status === 504) {
         errText = '504 Gateway Timeout — o purge pode demorar vários minutos. Verifique na VM112 se concluiu antes de repetir.';
+      } else if (res.status === 409) {
+        errText = typeof detail === 'string'
+          ? detail
+          : 'Purge já em curso para este domínio — aguarde a conclusão no drawer lateral.';
+      } else if (res.status === 503) {
+        errText = typeof detail === 'string'
+          ? detail
+          : 'Servidor ocupado — aguarde alguns segundos e tente novamente.';
+      } else if (res.status === 500 && errText === '500') {
+        errText = 'Erro interno ao iniciar purge — verifique logs da API Desk (VM122).';
       }
       throw new Error(errText);
     }