Disable target paging after hop login and keep CLI echo on timeouts.

LLDP discover and config collect were timing out on --More--; send vendor paging-off after nested/bastion auth, and attach session-log tails so failures show where the CLI stuck.

Co-authored-by: Cursor <cursoragent@cursor.com>

netx_api/config_sync_runner.py

@@ -35,7 +35,7 @@ def _utcnow() -> datetime:
 
 
 def _format_error(exc: BaseException) -> str:
-    return format_cli_failure(exc, limit=1020)
+    return format_cli_failure(exc, limit=4000)
 
 
 def _pool_for_cycle(cycle_id: str, concurrency: int) -> ThreadPoolExecutor:
@@ -124,16 +124,27 @@ def _collect_commands(
     *,
     conn_holder: dict[str, Any] | None = None,
 ) -> list[str]:
+    from .ne_netmiko import disable_target_paging
+
     per_cmd = int(settings.ne_collect_read_timeout_sec or 120)
     session_timeout = per_cmd * max(1, len(commands)) + 60
     log_buf = io.BytesIO()
     try:
         conn = open_netmiko_connection(creds, session_timeout=session_timeout, session_log=log_buf)
     except Exception as exc:
-        raise RuntimeError(format_cli_failure(exc, session_log_text(log_buf))) from exc
+        raise RuntimeError(format_cli_failure(exc, session_log_text(log_buf), limit=4000)) from exc
     if conn_holder is not None:
         conn_holder["conn"] = conn
+        conn_holder["session_log"] = log_buf
     try:
+        try:
+            disable_target_paging(
+                conn,
+                vendor=str(creds.get("vendor") or ""),
+                device_type=str(creds.get("device_type") or ""),
+            )
+        except Exception:
+            pass
         outputs: list[str] = []
         for command in commands:
             if conn_holder is not None and conn_holder.get("timed_out"):
@@ -141,9 +152,13 @@ def _collect_commands(
             try:
                 out = send_show_command(conn, command, read_timeout=per_cmd)
             except Exception as exc:
-                raise RuntimeError(format_cli_failure(exc, session_log_text(log_buf))) from exc
+                raise RuntimeError(format_cli_failure(exc, session_log_text(log_buf), limit=4000)) from exc
             outputs.append(str(out or ""))
         return outputs
+    except Exception as exc:
+        if isinstance(exc, TimeoutError):
+            raise RuntimeError(format_cli_failure(exc, session_log_text(log_buf), limit=4000)) from exc
+        raise
     finally:
         if conn_holder is not None:
             conn_holder.pop("conn", None)
@@ -157,13 +172,18 @@ def _collect_with_timeout(creds: dict[str, Any], commands: list[str]) -> list[st
     cap = int(settings.ne_collect_run_timeout_cap_sec or 600)
     budget = min(cap, per_cmd * max(1, len(commands)) + 90)
     holder: dict[str, Any] = {}
-    return run_cli_with_timeout(
-        lambda: _collect_commands(creds, commands, conn_holder=holder),
-        timeout_sec=budget,
-        conn_holder=holder,
-        label="config_sync",
-        acquire_budget=True,
-    )
+    try:
+        return run_cli_with_timeout(
+            lambda: _collect_commands(creds, commands, conn_holder=holder),
+            timeout_sec=budget,
+            conn_holder=holder,
+            label="config_sync",
+            acquire_budget=True,
+        )
+    except TimeoutError as exc:
+        raise RuntimeError(
+            format_cli_failure(exc, session_log_text(holder.get("session_log")), limit=4000)
+        ) from exc
 
 
 def _history_keep(db) -> int:

netx_api/ne_cli_errors.py

@@ -89,7 +89,17 @@ def format_cli_failure(exc: BaseException | str, transcript: str = "", *, limit:
         else:
             msg = f"auth_rejected: {auth}"
         return msg[:limit]
-    return exc_text[:limit]
+    # Non-auth: keep a session-log tail so operators can see where the CLI stuck
+    # (More prompt, half-auth, wrong command echo, etc.).
+    tail = str(transcript or "").strip()
+    if not tail:
+        return exc_text[:limit]
+    sep = "\n--- session log ---\n"
+    budget = max(120, int(limit) - len(exc_text) - len(sep) - 8)
+    clipped = tail[-budget:]
+    if len(tail) > budget:
+        clipped = "…\n" + clipped
+    return f"{exc_text}{sep}{clipped}"[:limit]
 
 
 def session_log_text(session_log: Any) -> str:

netx_api/ne_collect_runner.py

@@ -47,9 +47,12 @@ def shutdown_ne_collect_executor(*, wait: bool = False) -> None:
 
 def _format_run_error(exc: BaseException) -> str:
     head = f"{type(exc).__name__}: {exc}"
+    # RuntimeError from format_cli_failure already carries session log — keep it.
+    if isinstance(exc, RuntimeError) and "--- session log ---" in str(exc):
+        return str(exc)[:4000]
     tb = traceback.format_exc().strip()
     text = f"{head}\n{tb}" if tb else head
-    return text[:1020]
+    return text[:4000]
 
 
 def _safe_filename_part(text: str) -> str:
@@ -64,23 +67,61 @@ def _collect_on_device(
     read_timeout_sec: int | None = None,
     conn_holder: dict[str, Any] | None = None,
 ) -> str:
+    import io
+
+    from .ne_cli_errors import format_cli_failure, session_log_text
+    from .ne_netmiko import disable_target_paging
+
     per_cmd = int(read_timeout_sec if read_timeout_sec is not None else (settings.ne_collect_read_timeout_sec or 120))
     session_timeout = per_cmd * max(1, len(commands)) + 60
-    conn = open_netmiko_connection(creds, session_timeout=session_timeout)
-    if conn_holder is not None:
-        conn_holder["conn"] = conn
+    log_buf = io.BytesIO()
+    chunks: list[str] = []
+    conn = None
     try:
+        try:
+            conn = open_netmiko_connection(
+                creds, session_timeout=session_timeout, session_log=log_buf
+            )
+        except Exception as exc:
+            raise RuntimeError(format_cli_failure(exc, session_log_text(log_buf), limit=4000)) from exc
+        if conn_holder is not None:
+            conn_holder["conn"] = conn
+            conn_holder["session_log"] = log_buf
+        # Belt-and-suspenders: hop/bastion nested CLIs and missed session_prep.
+        try:
+            disable_target_paging(
+                conn,
+                vendor=str(creds.get("vendor") or ""),
+                device_type=str(creds.get("device_type") or ""),
+            )
+        except Exception:
+            _log.debug("collection paging disable failed", exc_info=True)
         prompt = str(conn.find_prompt() or "")
-        chunks: list[str] = []
         for command in commands:
             if conn_holder is not None and conn_holder.get("timed_out"):
                 raise TimeoutError("collection_aborted")
             ts = datetime.now().isoformat(timespec="seconds")
             chunks.append(f'>>> [{ts}] {{"String":"{command}", "Match":"{prompt}", "Timeout":0}}\n')
-            out = send_show_command(conn, command, read_timeout=per_cmd)
+            try:
+                out = send_show_command(conn, command, read_timeout=per_cmd)
+            except Exception as exc:
+                partial = "".join(chunks)
+                transcript = session_log_text(log_buf) or partial
+                raise RuntimeError(
+                    format_cli_failure(exc, transcript, limit=4000)
+                ) from exc
             chunks.append(str(out or ""))
             chunks.append("\n")
         return "".join(chunks)
+    except Exception as exc:
+        # Surface echo for mid-command Netmiko failures not already wrapped.
+        if isinstance(exc, RuntimeError) and "--- session log ---" in str(exc):
+            raise
+        partial = "".join(chunks)
+        transcript = session_log_text(log_buf) or partial
+        if transcript:
+            raise RuntimeError(format_cli_failure(exc, transcript, limit=4000)) from exc
+        raise
     finally:
         if conn_holder is not None:
             conn_holder.pop("conn", None)
@@ -94,20 +135,25 @@ def _collect_with_timeout(
     read_timeout_sec: int | None = None,
 ) -> str:
     from .cli_timeout import run_cli_with_timeout
+    from .ne_cli_errors import format_cli_failure, session_log_text
 
     per_cmd = int(read_timeout_sec if read_timeout_sec is not None else (settings.ne_collect_read_timeout_sec or 120))
     cap = int(settings.ne_collect_run_timeout_cap_sec or 600)
     budget = min(cap, per_cmd * max(1, len(commands)) + 90)
     holder: dict[str, Any] = {}
-    return run_cli_with_timeout(
-        lambda: _collect_on_device(
-            creds, commands, read_timeout_sec=per_cmd, conn_holder=holder
-        ),
-        timeout_sec=budget,
-        conn_holder=holder,
-        label="collection",
-        acquire_budget=True,
-    )
+    try:
+        return run_cli_with_timeout(
+            lambda: _collect_on_device(
+                creds, commands, read_timeout_sec=per_cmd, conn_holder=holder
+            ),
+            timeout_sec=budget,
+            conn_holder=holder,
+            label="collection",
+            acquire_budget=True,
+        )
+    except TimeoutError as exc:
+        transcript = session_log_text(holder.get("session_log"))
+        raise RuntimeError(format_cli_failure(exc, transcript, limit=4000)) from exc
 
 
 def _update_run(run_id: str, **fields: Any) -> None:

netx_api/ne_exec.py

@@ -81,13 +81,16 @@ def execute_managed_ne_commands(
     try:
         output = _collect_on_device(creds, cmds, read_timeout_sec=read_timeout)
     except Exception as exc:
+        detail = str(exc)[:4000]
         return {
             "ok": False,
             "device": device,
             "commands": cmds,
             "read_timeout_sec": read_timeout,
             "error": type(exc).__name__,
-            "detail": str(exc)[:2000],
+            "detail": detail,
+            # Preserve whatever echo was captured so LLDP discover / ops can show it.
+            "output": detail,
         }
 
     if len(output) > _EXEC_MAX_OUTPUT:

netx_api/ne_netmiko.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import time
 from typing import Any
 
 
@@ -37,6 +38,63 @@ def is_zte_device_type(device_type: str) -> bool:
     return "zte" in str(device_type or "").strip().lower()
 
 
+def is_huawei_device_type(device_type: str) -> bool:
+    return "huawei" in str(device_type or "").strip().lower()
+
+
+def paging_disable_commands(*, vendor: str = "", device_type: str = "") -> list[str]:
+    """Vendor CLI commands to disable --More-- paging (collection / LLDP / sync)."""
+    dt = str(device_type or "").strip().lower()
+    v = str(vendor or "").strip().lower()
+    blob = f"{dt} {v}"
+    if "huawei" in blob or "华为" in v:
+        return ["screen-length 0 temporary"]
+    if "hp_comware" in blob or "h3c" in blob or "comware" in blob:
+        return ["screen-length disable"]
+    if "juniper" in blob or "junos" in blob:
+        return ["set cli screen-length 0"]
+    if "nokia" in blob or "alcatel_sros" in blob or "sros" in blob:
+        return ["environment no more"]
+    if "alcatel_aos" in blob:
+        return ["terminal length 0"]
+    # Cisco / ZTE / generic SSH CLIs
+    return ["terminal length 0"]
+
+
+def disable_target_paging(
+    conn: Any,
+    *,
+    vendor: str = "",
+    device_type: str = "",
+) -> str:
+    """Send paging-off on the *current* CLI (critical after CLI hop / bastion jump).
+
+    Nested stelnet/telnet lands on the target without Netmiko ``session_preparation``,
+    so --More-- stays enabled and long ``display/show lldp`` / config dumps time out.
+    Uses timing reads so a wrong hop ``base_prompt`` does not block.
+    """
+    cmds = paging_disable_commands(vendor=vendor, device_type=device_type)
+    chunks: list[str] = []
+    for cmd in cmds:
+        try:
+            out = conn.send_command_timing(cmd, read_timeout=15)
+            chunks.append(str(out or ""))
+            continue
+        except Exception:
+            pass
+        try:
+            ret = getattr(conn, "RETURN", None) or "\n"
+            conn.write_channel(str(cmd) + ret)
+            time.sleep(0.35)
+            if hasattr(conn, "read_channel"):
+                part = conn.read_channel()
+                if part:
+                    chunks.append(str(part))
+        except Exception:
+            pass
+    return "".join(chunks)
+
+
 def send_show_command(conn: Any, command: str, *, read_timeout: int = 120) -> str:
     """Send a show/display command via ``send_command`` (wait for device prompt).
 

netx_api/ne_session_connect.py

@@ -360,14 +360,60 @@ def _zte_collection_driver_class(base_cls: type) -> type:
     return _ZteCollectionSession
 
 
+def _huawei_collection_driver_class(base_cls: type) -> type:
+    """Huawei VRP collection: ensure screen-length is off before long display cmds."""
+
+    class _HuaweiCollectionSession(base_cls):  # type: ignore[misc,valid-type]
+        def session_preparation(self) -> None:
+            prompt_pat = r"(?:<[^>\r\n]{1,64}>|\[[^\]\r\n]{1,64}\])"
+            try:
+                self._test_channel_read(pattern=prompt_pat)
+            except Exception:
+                try:
+                    self.write_channel(self.RETURN)
+                except Exception:
+                    pass
+                try:
+                    self._test_channel_read(pattern=prompt_pat)
+                except Exception:
+                    # Fall back to stock prompt family.
+                    self._test_channel_read(pattern=r"[>\]]")
+            try:
+                self.set_base_prompt()
+            except Exception:
+                pass
+            try:
+                self.disable_paging(
+                    command="screen-length 0 temporary",
+                    cmd_verify=False,
+                    pattern=prompt_pat,
+                )
+            except Exception:
+                try:
+                    self.send_command_timing("screen-length 0 temporary", read_timeout=15)
+                except Exception:
+                    pass
+            try:
+                self.clear_buffer()
+            except Exception:
+                pass
+
+    _HuaweiCollectionSession.__name__ = (
+        f"HuaweiCollection{getattr(base_cls, '__name__', 'Netmiko')}"
+    )
+    return _HuaweiCollectionSession
+
+
 def _collection_driver_class(device_type: str, base_cls: type) -> type:
     """Vendor-specific collection session prep (non-interactive CLI / LLDP / sync)."""
-    from .ne_netmiko import is_cisco_ios_device_type, is_zte_device_type
+    from .ne_netmiko import is_cisco_ios_device_type, is_huawei_device_type, is_zte_device_type
 
     if is_cisco_ios_device_type(device_type):
         return _cisco_ios_collection_driver_class(base_cls)
     if is_zte_device_type(device_type):
         return _zte_collection_driver_class(base_cls)
+    if is_huawei_device_type(device_type):
+        return _huawei_collection_driver_class(base_cls)
     return base_cls
 
 
@@ -910,6 +956,20 @@ def _connect_via_cli_hop(
             progress_cb=progress_cb,
             emit_raw=emit_raw,
         )
+        # Nested target CLI never got Netmiko session_preparation — turn off paging now.
+        if not interactive:
+            from .ne_netmiko import disable_target_paging
+
+            try:
+                paging_out = disable_target_paging(
+                    conn,
+                    vendor=str(creds.get("vendor") or ""),
+                    device_type=str(creds.get("device_type") or ""),
+                )
+                if paging_out and emit_raw:
+                    _emit_progress(progress_cb, paging_out)
+            except Exception:
+                _log.debug("cli hop target paging disable failed", exc_info=True)
         _attach_cli_hop_guard(
             conn,
             hop_prompt=hop_prompt,
@@ -1060,6 +1120,21 @@ def _connect_via_bastion(
         except Exception:
             pass
         raise
+    # After secondary auth (or bastion-managed landing), ensure target paging is off.
+    # session_preparation may have run too early against proxy banners.
+    if not interactive:
+        from .ne_netmiko import disable_target_paging
+
+        try:
+            paging_out = disable_target_paging(
+                conn,
+                vendor=str(creds.get("vendor") or ""),
+                device_type=str(creds.get("device_type") or ""),
+            )
+            if paging_out and not teed:
+                _emit_progress(progress_cb, paging_out)
+        except Exception:
+            _log.debug("bastion target paging disable failed", exc_info=True)
     return conn
 
 

netx_api/port_traffic_runner.py

@@ -260,10 +260,20 @@ def _sample_targets_shared_session(device_id: str, target_ids: list[str]) -> tup
     holder: dict[str, Any] = {}
 
     def _run_session() -> tuple[int, str]:
+        from .ne_netmiko import disable_target_paging
+
         conn = open_netmiko_connection(creds, session_timeout=budget)
         holder["conn"] = conn
         local_errors = 0
         try:
+            try:
+                disable_target_paging(
+                    conn,
+                    vendor=str(creds.get("vendor") or ""),
+                    device_type=str(creds.get("device_type") or ""),
+                )
+            except Exception:
+                pass
             for tid, ifname in ifaces:
                 if holder.get("timed_out"):
                     raise TimeoutError("port_traffic_aborted")

netx_api/topology_discover_scan.py

@@ -134,12 +134,16 @@ def _discover_one_target(
                 "error": detail,
             }
         if not exec_out.get("ok"):
-            err = str(exec_out.get("error") or exec_out.get("detail") or "exec_failed")[:500]
+            raw = str(exec_out.get("output") or exec_out.get("detail") or "")
+            err = str(exec_out.get("detail") or exec_out.get("error") or "exec_failed")[:4000]
             return {
                 **base,
                 "ok": False,
                 "command": cmd,
+                "parser_key": pkey,
+                "parser_stub": bool(is_stub),
                 "error": err,
+                "raw_preview": _raw_preview(raw or err),
             }
 
         raw = str(exec_out.get("output") or "")