changing default for log rotation

ssh_honeypot.py

@@ -38,23 +38,43 @@ DEFAULT_BANNER = "SSH-2.0-OpenSSH_8.9p1 Ubuntu-3ubuntu0.4"
 _event_logger = logging.getLogger("honeypot.events")    # -> human-readable
 _json_logger = logging.getLogger("honeypot.json")       # -> JSON Lines
 
+# Size at which a log file rotates — only applies when rotation is enabled
+# (see _log_handler).
+LOG_MAX_BYTES = 100 * 1024 * 1024
 
-def setup_logging(log_dir: str) -> None:
+
+def _log_handler(path: str, backup_count: int) -> RotatingFileHandler:
+    """File handler for one log file; ``backup_count <= 0`` disables rotation.
+
+    Rotation only ever *prunes* — the oldest segment is deleted to make room.
+    For a honeypot that is the wrong default: losing captured attempts is worse
+    than a large file. So unless a backup count is configured we pass
+    ``maxBytes=0``, which makes RotatingFileHandler never roll over, and the
+    file simply grows. Watch disk usage if you leave it that way.
+    """
+    return RotatingFileHandler(
+        path,
+        maxBytes=LOG_MAX_BYTES if backup_count > 0 else 0,
+        backupCount=max(backup_count, 0),
+        encoding="utf-8",
+    )
+
+
+def setup_logging(log_dir: str, backup_count: int = 0) -> None:
     """Configure three sinks: JSON Lines file, text file, and stdout.
 
     Both ``_event_logger`` and ``_json_logger`` are configured here. They do
     not propagate to the root logger so paramiko's transport chatter stays out
     of our event logs.
+
+    ``backup_count`` is how many rotated segments to keep per file; 0 (the
+    default) means never rotate. The console copy is never rotated — that is
+    the container runtime's job.
     """
     os.makedirs(log_dir, exist_ok=True)
 
     # JSON Lines: one json.dumps() record per line, no extra formatting.
-    json_handler = RotatingFileHandler(
-        os.path.join(log_dir, "honeypot.jsonl"),
-        maxBytes=10 * 1024 * 1024,
-        backupCount=5,
-        encoding="utf-8",
-    )
+    json_handler = _log_handler(os.path.join(log_dir, "honeypot.jsonl"), backup_count)
     json_handler.setFormatter(logging.Formatter("%(message)s"))
     _json_logger.addHandler(json_handler)
     _json_logger.setLevel(logging.INFO)
@@ -64,12 +84,7 @@ def setup_logging(log_dir: str) -> None:
     # Local-time, space-separated, no timezone clutter — easy to scan in a terminal.
     text_fmt = logging.Formatter("%(asctime)s  %(message)s", "%Y-%m-%d %H:%M:%S")
 
-    text_handler = RotatingFileHandler(
-        os.path.join(log_dir, "honeypot.log"),
-        maxBytes=10 * 1024 * 1024,
-        backupCount=5,
-        encoding="utf-8",
-    )
+    text_handler = _log_handler(os.path.join(log_dir, "honeypot.log"), backup_count)
     text_handler.setFormatter(text_fmt)
 
     stdout_handler = logging.StreamHandler()  # defaults to stderr/stdout stream
@@ -80,8 +95,15 @@ def setup_logging(log_dir: str) -> None:
     _event_logger.setLevel(logging.INFO)
     _event_logger.propagate = False
 
-    # Keep paramiko's internal logging quiet but capture genuine errors.
-    logging.getLogger("paramiko").setLevel(logging.WARNING)
+    # Silence paramiko's own logging. An internet-facing honeypot constantly
+    # gets half-open / non-SSH probes (port scanners, nc, health checks) that
+    # make paramiko emit noisy "Error reading SSH protocol banner" tracebacks.
+    # We already record a clean `connection_error` line for these ourselves, so
+    # raise paramiko above ERROR and give it a NullHandler so nothing falls
+    # through to Python's last-resort stderr handler.
+    paramiko_log = logging.getLogger("paramiko")
+    paramiko_log.setLevel(logging.CRITICAL)
+    paramiko_log.addHandler(logging.NullHandler())
 
 
 def _sanitize(value) -> str:
@@ -105,7 +127,10 @@ def _human_message(event: str, fields: dict) -> str:
     fields that matter) instead of a generic key=value dump, so a stream of
     attempts is easy to read at a glance.
     """
-    src = f"{fields['src_ip']}:{fields.get('src_port', '?')}" if "src_ip" in fields else ""
+    # Source IP and port as two separate aligned columns (IPv4 fits in 15, port
+    # in 5). They are already separate fields in the JSON Lines output.
+    src = (f"ip={fields['src_ip']:<15} port={fields.get('src_port', '?')!s:<5}"
+           if "src_ip" in fields else "")
 
     if event == "login_attempt":
         if fields.get("auth_method") == "password":
@@ -114,13 +139,13 @@ def _human_message(event: str, fields: dict) -> str:
             cred = (f"user={fields.get('username')!r} "
                     f"key={fields.get('key_type')}/{fields.get('key_fingerprint')}")
         client = _sanitize(fields.get("client_version", ""))
-        return f"{src:<21}  {cred}  [{fields.get('auth_method')}] {client}".rstrip()
+        return f"{src}  {cred}  [{fields.get('auth_method')}] {client}".rstrip()
     if event == "connection":
-        return f"{src:<21}  new connection"
+        return f"{src}  new connection"
     if event == "disconnect":
-        return f"{src:<21}  disconnected"
+        return f"{src}  disconnected"
     if event == "connection_error":
-        return f"{src:<21}  error: {_sanitize(fields.get('error'))}"
+        return f"{src}  error: {_sanitize(fields.get('error'))}"
     if event == "server_start":
         return (f"listening on {fields.get('bind')}:{fields.get('port')}  "
                 f"banner={fields.get('banner')!r}  fingerprint={fields.get('host_key_fingerprint')}")
@@ -345,7 +370,10 @@ class Honeypot:
 # --------------------------------------------------------------------------- #
 
 def _env_default(name: str, fallback: str) -> str:
-    return os.environ.get(name, fallback)
+    # `or fallback` rather than a .get() default: a set-but-empty variable is
+    # common in quadlet/compose files and systemd unit drop-ins, and .get()
+    # would hand back "" — which then crashes int() for the numeric options.
+    return os.environ.get(name) or fallback
 
 
 def parse_args(argv=None) -> argparse.Namespace:
@@ -359,6 +387,11 @@ def parse_args(argv=None) -> argparse.Namespace:
                    help="Path to the persisted host key (env HONEYPOT_HOST_KEY)")
     p.add_argument("--log-dir", default=_env_default("HONEYPOT_LOG_DIR", "./logs"),
                    help="Directory for log files (env HONEYPOT_LOG_DIR)")
+    p.add_argument("--log-backups", type=int,
+                   default=int(_env_default("HONEYPOT_LOG_BACKUPS", "0")),
+                   help="Rotated log segments to keep per file, rotating every "
+                        f"{LOG_MAX_BYTES // (1024 * 1024)} MB. 0 (the default) never "
+                        "rotates and lets the files grow (env HONEYPOT_LOG_BACKUPS)")
     p.add_argument("--banner", default=_env_default("HONEYPOT_BANNER", DEFAULT_BANNER),
                    help="Spoofed SSH version string (env HONEYPOT_BANNER)")
     p.add_argument("--max-attempts", type=int,
@@ -372,7 +405,7 @@ def parse_args(argv=None) -> argparse.Namespace:
 
 def main(argv=None) -> int:
     args = parse_args(argv)
-    setup_logging(args.log_dir)
+    setup_logging(args.log_dir, args.log_backups)
 
     if not args.banner.startswith("SSH-2.0-"):
         log_event("config_error", error="banner must start with 'SSH-2.0-'")