fix(snmp): bundle pysnmp in image and port poller to the asyncio HLAPI

The SNMP plugin ships in the image but logged "pysnmp not installed — SNMP
polling disabled" on every poll, so polling never worked. Two coupled defects:

1. The Dockerfile installed only `.[ansible]`, so the `snmp` extra (pysnmp)
   was never bundled even though the plugin is first-party and shipped.
2. poller.py used the synchronous pysnmp HLAPI (`next(getCmd(...))`), which
   pysnmp-lextudio 6.x removed — it's asyncio-only now — so even with the dep
   present, polling would have thrown and silently returned nothing. The 5.x
   line that still has the sync API isn't safe on the image's Python 3.13.

Fix:
- Dockerfile: install `.[ansible,snmp]`.
- poller.py: `poll_device_sync` → `async def poll_device` on the asyncio HLAPI,
  with a dual-version import (pysnmp 7.x `pysnmp.hlapi.v3arch.asyncio`/`get_cmd`
  + async `UdpTransportTarget.create`; pysnmp-lextudio 6.2.x
  `pysnmp.hlapi.asyncio`/`getCmd` + direct `UdpTransportTarget`) so a dependency
  bump can't silently re-break it.
- scheduler.py: await poll_device directly; drop the run_in_executor wrapper
  and the now-unused asyncio import.
- Add tests/plugins/snmp/test_poller.py covering the version→mpModel mapping,
  that the poller is a coroutine, and the graceful no-pysnmp path.

Note: CI confirms import/load and the no-pysnmp path, but has no SNMP target —
live polling against real devices is verified after deploy.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016Jg27rgypiW2efULXJDtMC

plugins/snmp/poller.py

@@ -1,11 +1,13 @@
 # plugins/snmp/poller.py
 """
-Synchronous SNMP GET helper, run via executor.
+Asynchronous SNMP GET helper.
 
-Requires pysnmp-lextudio (maintained pysnmp fork):
+Requires pysnmp-lextudio (the maintained pysnmp fork), bundled into the Docker
+image via the `snmp` extra (`pip install .[snmp]`):
     pip install 'steward[snmp]'
 
-If pysnmp is not installed, poll_device() returns an empty dict and logs a warning.
+If pysnmp is not installed, poll_device() returns an empty dict and logs a
+warning — SNMP polling is then simply disabled, nothing else breaks.
 """
 from __future__ import annotations
 import logging
@@ -22,39 +24,74 @@ def _pysnmp_available() -> bool:
 
 
 def _mp_model(version: str) -> int:
-    """Map version string to pysnmp mpModel integer."""
+    """Map an SNMP version string to pysnmp's mpModel int (0 = v1, 1 = v2c)."""
     return 0 if version == "1" else 1
 
 
-def poll_device_sync(
+async def poll_device(
     host: str,
     port: int,
     community: str,
     version: str,
     oids: list[dict],
 ) -> dict[str, float]:
-    """
-    Perform SNMP GET for each OID and return {label: float_value}.
-    Non-numeric OIDs (strings, etc.) are skipped.
-    Returns empty dict on any error.
+    """Perform an SNMP GET for each OID and return ``{label: float_value}``.
+
+    Non-numeric OIDs (strings, etc.) are skipped. Returns an empty dict on any
+    error (unreachable host, wrong community, …) so a flaky device never breaks
+    the poll loop.
+
+    pysnmp's HLAPI is asyncio-only as of v6. The import path moved between major
+    versions, so we support both rather than let a dependency bump silently
+    re-break polling:
+      • pysnmp-lextudio 6.2.x → ``pysnmp.hlapi.asyncio`` (``getCmd`` + a directly
+        constructed ``UdpTransportTarget``).
+      • canonical pysnmp 7.x  → ``pysnmp.hlapi.v3arch.asyncio`` (``get_cmd`` + the
+        async ``UdpTransportTarget.create``).
     """
     if not _pysnmp_available():
         logger.warning("pysnmp not installed — SNMP polling disabled. "
                        "Install with: pip install 'steward[snmp]'")
         return {}
 
-    from pysnmp.hlapi import (
-        CommunityData,
-        ContextData,
-        ObjectIdentity,
-        ObjectType,
-        SnmpEngine,
-        UdpTransportTarget,
-        getCmd,
-    )
+    try:
+        # canonical pysnmp 7.x
+        from pysnmp.hlapi.v3arch.asyncio import (
+            CommunityData,
+            ContextData,
+            ObjectIdentity,
+            ObjectType,
+            SnmpEngine,
+            UdpTransportTarget,
+            get_cmd as _get_cmd,
+        )
+        _transport_is_async = True
+    except ImportError:
+        # pysnmp-lextudio 6.2.x
+        from pysnmp.hlapi.asyncio import (
+            CommunityData,
+            ContextData,
+            ObjectIdentity,
+            ObjectType,
+            SnmpEngine,
+            UdpTransportTarget,
+            getCmd as _get_cmd,
+        )
+        _transport_is_async = False
+
+    engine = SnmpEngine()
+
+    # Same host/port for every OID on this device, so build the transport once.
+    try:
+        if _transport_is_async:
+            transport = await UdpTransportTarget.create((host, port), timeout=5, retries=1)
+        else:
+            transport = UdpTransportTarget((host, port), timeout=5, retries=1)
+    except Exception as exc:
+        logger.debug("SNMP transport setup failed for %s:%s: %s", host, port, exc)
+        return {}
 
     results: dict[str, float] = {}
-    engine = SnmpEngine()
 
     for oid_cfg in oids:
         oid = oid_cfg["oid"]
@@ -62,14 +99,12 @@ def poll_device_sync(
         scale = float(oid_cfg.get("scale", 1.0))
 
         try:
-            error_indication, error_status, error_index, var_binds = next(
-                getCmd(
-                    engine,
-                    CommunityData(community, mpModel=_mp_model(version)),
-                    UdpTransportTarget((host, port), timeout=5, retries=1),
-                    ContextData(),
-                    ObjectType(ObjectIdentity(oid)),
-                )
+            error_indication, error_status, error_index, var_binds = await _get_cmd(
+                engine,
+                CommunityData(community, mpModel=_mp_model(version)),
+                transport,
+                ContextData(),
+                ObjectType(ObjectIdentity(oid)),
             )
         except Exception as exc:
             logger.debug("SNMP GET %s@%s OID %s failed: %s", host, port, oid, exc)
@@ -88,7 +123,7 @@ def poll_device_sync(
             try:
                 results[label] = float(val) * scale
             except (TypeError, ValueError):
-                # Non-numeric type (e.g. OctetString description) — skip
+                # Non-numeric type (e.g. OctetString description) — skip.
                 logger.debug("SNMP non-numeric value for %s label=%s: %r", oid, label, val)
 
     return results

plugins/snmp/scheduler.py

@@ -1,6 +1,5 @@
 # plugins/snmp/scheduler.py
 from __future__ import annotations
-import asyncio
 import logging
 from typing import TYPE_CHECKING
 
@@ -27,15 +26,13 @@ def make_poll_task(app: "Quart") -> ScheduledTask:
 
 
 async def _do_poll(app: "Quart") -> None:
-    from .poller import poll_device_sync
+    from .poller import poll_device
     from steward.core.alerts import record_metric
 
     devices: list[dict] = app.config["PLUGINS"]["snmp"].get("devices", [])
     if not devices:
         return
 
-    loop = asyncio.get_event_loop()
-
     async with app.db_sessionmaker() as session:
         async with session.begin():
             for device in devices:
@@ -52,11 +49,7 @@ async def _do_poll(app: "Quart") -> None:
                     continue
 
                 try:
-                    readings = await loop.run_in_executor(
-                        None,
-                        poll_device_sync,
-                        host, port, community, version, oids,
-                    )
+                    readings = await poll_device(host, port, community, version, oids)
                 except Exception:
                     logger.exception("SNMP poll failed for device %s (%s)", name, host)
                     continue