fix: adopt rotated database leases

Assistant: codex
Assistant-Model: gpt-5.6-sol
Assistant-Session: 01a02b22-9638-76d2-bbff-b7ea1770b118

README.md

@@ -22,10 +22,14 @@ SQLite through `SBOM_NEXUS_DATABASE_PATH`; production uses
 `SBOM_NEXUS_DATABASE_URL_FILE=/var/run/secrets/.../url` and `make migrate`.
 The direct `SBOM_NEXUS_DATABASE_URL` variable remains available for disposable
 development environments; mounted secret files are preferred for production.
+The runtime rereads the mounted file for every new database connection and
+recycles pooled connections every 300 seconds by default, configurable with
+`SBOM_NEXUS_DATABASE_POOL_RECYCLE_SECONDS`.
 
 ## Initial API surface
 
 - `GET /state/health`
+- `GET /state/live`
 - `PUT /repositories/{repo_slug}`
 - `GET /sbom/catch-up?limit=3`
 - `POST /sbom/{repo_slug}/ingest`

src/sbom_nexus/api.py

@@ -207,9 +207,17 @@ def create_app(database_path: str | Path | None = None) -> FastAPI:
         version="0.1.0",
         description="SBOM capture, history, evaluation, and bounded catch-up service",
     )
-    application.state.store = Store(
+    configured_target = (
         database_path if database_path is not None else database_target("sbom-nexus.db")
     )
+    application.state.store = Store(
+        configured_target,
+        database_url_file=(
+            os.getenv("SBOM_NEXUS_DATABASE_URL_FILE")
+            if database_path is None
+            else None
+        ),
+    )
     if _auto_create(application.state.store):
         application.state.store.init_schema()
 
@@ -219,6 +227,10 @@ def create_app(database_path: str | Path | None = None) -> FastAPI:
         store.health()
         return {"status": "ok", "store": "connected", "dialect": store.dialect}
 
+    @application.get("/state/live")
+    def live() -> dict[str, str]:
+        return {"status": "ok"}
+
     @application.put("/repositories/{repo_slug}")
     def upsert_repository(
         repo_slug: str, body: RepositoryUpsert, request: Request

src/sbom_nexus/database.py

@@ -2,8 +2,10 @@
 
 from __future__ import annotations
 
+import os
 from pathlib import Path
 
+import psycopg
 from sqlalchemy import (
     JSON,
     Boolean,
@@ -128,9 +130,49 @@ def database_url(value: str | Path) -> str:
     return text
 
 
-def create_database_engine(value: str | Path) -> Engine:
+def _dynamic_connection_factory(url_file: Path):
+    def connect():
+        try:
+            value = url_file.read_text(encoding="utf-8").strip()
+        except OSError as exc:
+            raise RuntimeError("Unable to read mounted database URL") from exc
+        if not value:
+            raise RuntimeError("Mounted database URL is empty")
+        url = database_url(value)
+        if not url.startswith("postgresql+psycopg://"):
+            raise RuntimeError("Mounted database URL must use PostgreSQL")
+        return psycopg.connect(
+            url.replace("postgresql+psycopg://", "postgresql://", 1)
+        )
+
+    return connect
+
+
+def create_database_engine(
+    value: str | Path, *, database_url_file: str | Path | None = None
+) -> Engine:
     url = database_url(value)
     options: dict[str, object] = {"pool_pre_ping": True}
+    if database_url_file:
+        if not url.startswith("postgresql+psycopg://"):
+            raise RuntimeError("Dynamic database URL files require PostgreSQL")
+        try:
+            recycle_seconds = int(
+                os.getenv("SBOM_NEXUS_DATABASE_POOL_RECYCLE_SECONDS", "300")
+            )
+        except ValueError as exc:
+            raise RuntimeError(
+                "SBOM_NEXUS_DATABASE_POOL_RECYCLE_SECONDS must be an integer"
+            ) from exc
+        if recycle_seconds < 1:
+            raise RuntimeError(
+                "SBOM_NEXUS_DATABASE_POOL_RECYCLE_SECONDS must be positive"
+            )
+        options["creator"] = _dynamic_connection_factory(Path(database_url_file))
+        options["pool_recycle"] = recycle_seconds
+        # Keep only the dialect in SQLAlchemy's URL; the creator rereads the
+        # mounted value without retaining credentials in engine repr/logging.
+        url = "postgresql+psycopg://"
     if url.startswith("sqlite:"):
         options["connect_args"] = {"check_same_thread": False}
     engine = create_engine(url, **options)

src/sbom_nexus/storage.py

@@ -50,8 +50,15 @@ def datetime_text(value: datetime | str | None) -> str | None:
 
 
 class Store:
-    def __init__(self, database_target: str | Path) -> None:
-        self.engine = create_database_engine(database_target)
+    def __init__(
+        self,
+        database_target: str | Path,
+        *,
+        database_url_file: str | Path | None = None,
+    ) -> None:
+        self.engine = create_database_engine(
+            database_target, database_url_file=database_url_file
+        )
 
     @property
     def dialect(self) -> str:

tests/test_api.py

@@ -43,6 +43,7 @@ def test_health_and_legacy_ingest_query_and_licence_report(tmp_path: Path) -> No
         "store": "connected",
         "dialect": "sqlite",
     }
+    assert client.get("/state/live").json() == {"status": "ok"}
     register(client, "demo")
 
     response = client.post(

tests/test_database_rotation.py

@@ -0,0 +1,42 @@
+from __future__ import annotations
+
+from pathlib import Path
+
+from sbom_nexus import database
+
+
+def test_dynamic_connection_factory_rereads_mounted_url(monkeypatch, tmp_path: Path) -> None:
+    secret = tmp_path / "url"
+    observed: list[str] = []
+
+    def fake_connect(dsn: str):
+        observed.append(dsn)
+        return object()
+
+    monkeypatch.setattr(database.psycopg, "connect", fake_connect)
+    factory = database._dynamic_connection_factory(secret)
+    secret.write_text("postgresql://first@db/sbom\n", encoding="utf-8")
+    factory()
+    secret.write_text("postgresql://second@db/sbom\n", encoding="utf-8")
+    factory()
+
+    assert observed == [
+        "postgresql://first@db/sbom",
+        "postgresql://second@db/sbom",
+    ]
+
+
+def test_dynamic_engine_keeps_credential_out_of_engine_url(
+    monkeypatch, tmp_path: Path
+) -> None:
+    secret = tmp_path / "url"
+    secret.write_text("postgresql://username:password@db/sbom\n", encoding="utf-8")
+    monkeypatch.setenv("SBOM_NEXUS_DATABASE_POOL_RECYCLE_SECONDS", "60")
+
+    engine = database.create_database_engine(
+        secret.read_text().strip(), database_url_file=secret
+    )
+
+    assert "username" not in str(engine.url)
+    assert "password" not in str(engine.url)
+    assert engine.pool._recycle == 60

workplans/SBOM-WP-0004-database-lease-rotation.md

@@ -0,0 +1,63 @@
+---
+id: SBOM-WP-0004
+type: workplan
+title: "Adopt rotated database leases without liveness restarts"
+domain: infotech
+repo: sbom-nexus
+status: active
+owner: codex
+topic_slug: infotech
+created: "2026-08-23"
+updated: "2026-08-23"
+quality_dor: DoR-Ok
+quality_dor_at: "2026-08-23"
+quality_dor_by: codex
+quality_dor_note: "CUST-IN-0014 supplies exact 30-minute expiry/restart evidence, mounted Secret rotation behavior, owner boundaries, value-safety constraints, and live acceptance criteria."
+origin: residual
+origin_ref: CUST-IN-0014
+related:
+  - RAPP-SBOM-NEXUS-WP-0003
+---
+
+# Adopt rotated database leases without liveness restarts
+
+## Reread the mounted DSN on new pool connections
+
+```task
+id: SBOM-WP-0004-T01
+status: done
+priority: high
+```
+
+Use a value-safe SQLAlchemy connection creator that rereads the mounted URL
+for every new DBAPI connection. Recycle pooled connections before the current
+30-minute lease expires and preserve `pool_pre_ping` so revoked sessions are
+replaced with the current mounted credential.
+
+Completed with tests proving two connection attempts observe two file values
+and that the engine URL contains neither username nor password.
+
+## Separate liveness from database readiness
+
+```task
+id: SBOM-WP-0004-T02
+status: done
+priority: high
+```
+
+Expose process-only `/state/live` while retaining the database-backed
+`/state/health` readiness contract. The package must move only liveness to the
+new route so a transient credential handoff removes traffic but does not ask
+Kubernetes to restart an otherwise healthy process.
+
+## Prove one complete live lease rotation
+
+```task
+id: SBOM-WP-0004-T03
+status: wait
+priority: high
+```
+
+Promote the shared new image/package digest, observe at least one complete
+database lease rotation, and require continuous process uptime, recovered
+readiness, no liveness-driven restart, and no credential values in logs.