fix: the web service waits for the PostgreSQL driver instead of failing at start.

ADR.md

@@ -359,11 +359,16 @@ missing. PostgreSQL may also start after the services at boot.
 Decision: The web service makes its signing key and creates the refresh tokens' database itself,
 and the Router Manager its sessions database; neither `vpn-router-setup` nor a maintainer script
 creates them. Each service does it before it listens, retrying every 15 seconds and logging each
-failure, then listens.
+failure, then listens. The first step is the PostgreSQL driver: psycopg finds libpq through the
+linker cache, which the package manager refreshes only at the end of an installation, so a
+service started during the installation cannot import it yet. The Manager imports psycopg only in
+its retry loop. On the router, `vpnrouter_auth` alone imports it, in `driver()`, which the web
+service's retry loop calls first, and turns a failed import and every database failure into
+`DatabaseError`, the only database exception the web service and `vpn-router-user` handle.
 
-Consequences: The order of installation and of the start of PostgreSQL does not matter; a
+Consequences: The order of installation and of the start of PostgreSQL and of the linker cache
-database that cannot be created is created once the cause is gone, and the reason is in the
+does not matter; a driver that cannot be loaded or a database that cannot be created is retried
-service's log. The web unit may write `/etc/vpn-router` for the key and the database password.
+until the cause is gone, and the reason is in the service's log, without restarts. The web unit may write `/etc/vpn-router` for the key and the database password.
 
 ## ADR-17: Bootstrap diagnostics run only until the first setup succeeds
 

debian-package/debian/changelog

@@ -1,3 +1,16 @@
+vpn-router (0.9.1-1) unstable; urgency=medium
+
+  fix: the web service waits for the PostgreSQL driver instead of failing at start.
+
+  * The web service imports psycopg as the first step of its
+    initialization, retrying every 15 seconds and logging the failure while
+    libpq cannot be found yet during an installation.
+  * vpn-router-user and vpn-router-token load without psycopg; a command
+    that needs the token database then reports that the driver cannot be
+    loaded, as it reports every failure of that database.
+
+ -- Slawomir Koszewski <slawek@koszewscy.waw.pl>  Sun, 25 Oct 2026 05:00:00 +0200
+
 vpn-router (0.9.0-1) unstable; urgency=medium
 
   feature: bootstrap diagnostics on port 17081 during the first setup.

debian-package/src/usr/lib/vpn-router/vpnrouter_auth.py

@@ -5,12 +5,17 @@ router and that never leaves it. /auth issues an access token naming the user in
 a refresh token valid REFRESH_SECONDS; vpn-router-token issues tokens with the admin claim and
 any lifetime. A token without the admin claim must expire within MAX_LIFETIME seconds.
 
+psycopg is imported by driver() when the database is used, not when the module is loaded, so the
+module loads while libpq cannot be found yet. A driver that cannot be imported and every
+failure of the database raise DatabaseError, the only database exception callers see.
+
 Refresh tokens are random strings kept in PostgreSQL as their SHA-256, in the database of
 web.database, by default the vpn-router database on this host that provision() creates on the
 web service's first run, with the password in DATABASE_SECRET. A refresh token is used once: refreshing replaces it.
 """
 
 import base64
+import contextlib
 import hashlib
 import hmac
 import json
@@ -19,9 +24,6 @@ import secrets
 import subprocess
 import time
 
-import psycopg
-from psycopg.conninfo import make_conninfo
-
 WEB_KEY = pathlib.Path('/etc/vpn-router/web.key')
 DATABASE_SECRET = pathlib.Path('/etc/vpn-router/database.secret')
 
@@ -41,6 +43,10 @@ LEEWAY = 60
 SCHEMA = 'CREATE TABLE IF NOT EXISTS refresh_tokens (hash text PRIMARY KEY, name text NOT NULL, expires timestamptz NOT NULL)'
 
 
+class DatabaseError(Exception):
+    """The PostgreSQL driver cannot be loaded, or the database of the refresh tokens failed."""
+
+
 def key():
     """The signing key, or '' while the web service has not made it."""
     try:
@@ -111,20 +117,34 @@ def admin_token(seconds):
     return {'token': sign({'admin': True, 'iat': now, 'exp': now + seconds}, secret), 'expires': now + seconds}
 
 
+def driver():
+    """The psycopg module; DatabaseError while libpq cannot be found."""
+    try:
+        import psycopg
+    except ImportError as exc:
+        raise DatabaseError(f'the PostgreSQL driver cannot be loaded: {exc}') from exc
+    return psycopg
+
+
 def connection_string(cfg):
-    """web.database, or the database provision() creates."""
+    """web.database, or the database provision() creates; its password, of URL-safe characters, is written unquoted."""
     configured = cfg['web']['database'].strip()
     if configured:
         return configured
     password = DATABASE_SECRET.read_text(encoding='utf-8').strip()
-    return make_conninfo(host='localhost', dbname=DATABASE, user=DATABASE, password=password)
+    return f'host=localhost dbname={DATABASE} user={DATABASE} password={password}'
 
 
-def connect(cfg):
+@contextlib.contextmanager
-    """A connection to the database of the refresh tokens, with their table."""
+def database(cfg):
-    connection = psycopg.connect(connection_string(cfg), autocommit=True)
+    """A connection to the database of the refresh tokens, with their table, closed at the end; DatabaseError on any failure."""
+    psycopg = driver()
+    try:
+        with psycopg.connect(connection_string(cfg), autocommit=True) as connection:
             connection.execute(SCHEMA)
-    return connection
+            yield connection
+    except psycopg.Error as exc:
+        raise DatabaseError(f'the token database failed: {exc}') from exc
 
 
 def digest(token):
@@ -136,7 +156,7 @@ def issue(cfg, name):
     secret = key()
     now = int(time.time())
     refresh = secrets.token_urlsafe(32)
-    with connect(cfg) as connection:
+    with database(cfg) as connection:
         connection.execute('DELETE FROM refresh_tokens WHERE expires < now()')
         connection.execute("INSERT INTO refresh_tokens VALUES (%s, %s, now() + make_interval(secs => %s))",
                            (digest(refresh), name, REFRESH_SECONDS))
@@ -153,7 +173,7 @@ def take(cfg, token):
     """Remove an unexpired refresh token and return the user it was issued to, or None."""
     if not isinstance(token, str) or not token:
         return None
-    with connect(cfg) as connection:
+    with database(cfg) as connection:
         row = connection.execute('DELETE FROM refresh_tokens WHERE hash = %s RETURNING name, expires > now()',
                                  (digest(token),)).fetchone()
     return row[0] if row and row[1] else None
@@ -161,7 +181,7 @@ def take(cfg, token):
 
 def revoke(cfg, name):
     """Remove all refresh tokens of the user."""
-    with connect(cfg) as connection:
+    with database(cfg) as connection:
         connection.execute('DELETE FROM refresh_tokens WHERE name = %s', (name,))
 
 

debian-package/src/usr/lib/vpn-router/web

@@ -30,7 +30,9 @@ default the address the service listens on, is the last entry of its X-Forwarded
 Docker is installed, and the state of site-to-site, road-warrior (IPsec P2S), OpenVPN and
 WireGuard. No configuration data, key or address is shown.
 
-The service first initializes, without listening: it makes the signing key and the refresh
+The service first initializes, without listening: it loads the PostgreSQL driver
+(vpnrouter_auth.driver; psycopg finds libpq through the linker cache, which the package manager
+refreshes only at the end of an installation), then it makes the signing key and the refresh
 tokens' database (vpnrouter_auth), retrying every INIT_RETRY_SECONDS until both are ready and
 logging each failure. Then it listens.
 
@@ -58,7 +60,6 @@ from http.server import BaseHTTPRequestHandler
 sys.path.insert(0, os.path.dirname(os.path.realpath(__file__)))
 
 import ldap
-import psycopg
 
 import vpnrouter
 import vpnrouter_auth
@@ -440,8 +441,8 @@ def users(state, method, path, body):
         return USER_STATUS.get(failure.status, 500), {'error': str(failure)}
     except (ldap.LDAPError, OSError) as exc:
         return 500, {'error': f'the user directory failed: {exc}'}
-    except psycopg.Error as exc:
+    except vpnrouter_auth.DatabaseError as exc:
-        return 500, {'error': f'the token database failed: {exc}'}
+        return 500, {'error': str(exc)}
     except SystemExit as exc:
         # vpnrouter ends with SystemExit, carrying the reason, on a configuration it cannot read.
         return 500, {'error': str(exc.code)}
@@ -544,8 +545,8 @@ def auth(path, body):
         return (401 if failure.status == 3 else USER_STATUS.get(failure.status, 500)), {'error': str(failure)}
     except (ldap.LDAPError, OSError) as exc:
         return 500, {'error': f'the user directory failed: {exc}'}
-    except psycopg.Error as exc:
+    except vpnrouter_auth.DatabaseError as exc:
-        return 500, {'error': f'the token database failed: {exc}'}
+        return 500, {'error': str(exc)}
     except SystemExit as exc:
         return 500, {'error': str(exc.code)}
 
@@ -696,19 +697,22 @@ def log(message):
 
 
 def initialize():
-    """Make the signing key and the refresh tokens' database, retrying every INIT_RETRY_SECONDS until both are ready."""
+    """Load the PostgreSQL driver, make the signing key and the refresh tokens' database, retrying every INIT_RETRY_SECONDS until all are ready."""
     reported = None
     while True:
-        phase = 'reading the configuration'
+        phase = 'loading the PostgreSQL driver'
         try:
+            vpnrouter_auth.driver()
+            phase = 'reading the configuration'
             cfg = vpnrouter.load_config()
             phase = 'making the signing key'
             vpnrouter_auth.ensure_key()
             phase = 'creating the token database'
             vpnrouter_auth.provision(cfg)
             phase = 'connecting to the token database'
-            vpnrouter_auth.connect(cfg).close()
+            with vpnrouter_auth.database(cfg):
-        except (OSError, psycopg.Error, SystemExit) as exc:
+                pass
+        except (OSError, SystemExit, vpnrouter_auth.DatabaseError) as exc:
             error = f'{phase} failed: {exc.code if isinstance(exc, SystemExit) else exc}'
             if error != reported:
                 log(f'{error}; retrying every {INIT_RETRY_SECONDS} seconds')

debian-package/src/usr/sbin/vpn-router-user

@@ -29,8 +29,8 @@ import sys
 sys.path.insert(0, '/usr/lib/vpn-router')
 
 import ldap
-import psycopg
 
+import vpnrouter_auth
 import vpnrouter_users
 from vpnrouter_users import Failure
 
@@ -135,8 +135,8 @@ def main(argv):
         message, status = str(failure), failure.status
     except (ldap.LDAPError, OSError) as exc:
         message, status = f'the user directory failed: {exc}', 1
-    except psycopg.Error as exc:
+    except vpnrouter_auth.DatabaseError as exc:
-        message, status = f'the token database failed: {exc}', 1
+        message, status = str(exc), 1
     else:
         print(json.dumps(answer) if as_json else text_of(answer))
         return 0