Update README.md

Clarify that this agent is meant to run with Net SNMPd.

README.md

@@ -43,7 +43,10 @@ optional arguments:
 sudo cp dist/vpp-snmp-agent /usr/sbin/
 ```
 
-## Configuration file
+## Configuration
+
+This agent requires the `linux-cp` plugin to be enabled in VPP, and it requires read/write access
+to the VPP API and Stats sockets (typically in `/run/vpp/*.sock`).
 
 This SNMP Agent will read a [vppcfg](https://github.com/pimvanpelt/vppcfg) configuration file,
 which provides a mapping between VPP interface names, Linux Control Plane interface names, and
@@ -85,8 +88,12 @@ the `ifName`. However, if the config file is read, it will change the behavior a
 
 ## SNMPd config
 
-First, configure the snmpd to accept agentx connections by adding (at least) the following
+This agent is meant to run alongside the snmpd shipped in Debian (Bullseye or Bookworm), called
-to `snmpd.conf`:
+[Net SNMP](http://net-snmp.sourceforge.net/). The same snmpd is available in Ubuntu (Focal, Jammy) as well,
+which should work.
+
+After installing the snmpd (`apt install snmpd`), configure it to accept agentx connections by adding (at least)
+the following to `snmpd.conf`:
 ```
 master  agentx
 agentXSocket tcp:localhost:705,unix:/var/agentx-dataplane/master
@@ -119,15 +126,18 @@ sudo systemctl start vpp-snmp-agent
 
 # Support
 
-Limnited support is offered on this codebase. GitHub issues can be files for issues with the design or
+This software is compatible only with the current production release of VPP, which can be found on its
-implementation (eg bugs, feature requests), but no _user_ support can be given. Put simply, this repo
+[Gerrit](https://gerrit.fd.io/r/q/repo:vpp) service. Maintaining backwards compatibility is not a goal of
-accepts only bugreports with the code, not with its use.
+this repository.
+
+Limited support is offered on the codebase: GitHub issues may be filed for issues with the _design or
+implementation_ (eg. bugs, feature requests), but _user_ support can not be given. Put simply, this repo
+accepts only bugreports with the code, not with its use. See the LICENSE for clarity.
 
 Issues with the codebase that are well researched ([this article](https://marker.io/blog/how-to-write-bug-report)
 gives a good example of the expectation), preferably pointing at the location where the problem occurred,
 and if possible proposing a fix, are most welcome.
 
-Requests that are not discussing problems with the software itself, notably enduser support requests,
+Requests that don't discuss problems with the software itself, notably enduser support requests, will not be
-will not be handled unless they clearly demonstrate a bug and propose workarounds or fixes. 
+handled unless they clearly demonstrate a bug and propose workarounds or fixes. Paid support can be obtained
-
+on hourly commission. Reach out to IPng Networks GmbH (sales@ipng.ch) to discuss rates. 
-Paid support can be obtained on hourly commission. Reach out \to IPng Networks (sales@ipng.ch) to discuss rates. 

agentx/agent.py

@@ -28,7 +28,11 @@ class Agent(object):
         self._lastupdate = 0
         self._update_period = period  # Seconds
 
-        self._net = Network(server_address=server_address)
+        try:
+            debug = args.debug_agent
+        except:
+            debug = False
+        self._net = Network(server_address=server_address, debug=debug)
 
         self._oid_list = []
         self._args = args
@@ -67,10 +71,9 @@ class Agent(object):
             try:
                 self._net.run()
             except Exception as e:
-                self.logger.error("An exception occurred: %s" % e)
+                self.logger.error("Disconnecting due to exception: %s" % e)
-                self.logger.error("Reconnecting")
                 self._net.disconnect()
-                time.sleep(0.1)
+                time.sleep(1)
 
     def stop(self):
         self.logger.debug("Stopping")

agentx/network.py

@@ -27,11 +27,11 @@ class NetworkError(Exception):
 
 
 class Network:
-    def __init__(self, server_address="/var/agentx/master"):
+    def __init__(self, server_address="/var/agentx/master", debug=False):
 
         self.session_id = 0
         self.transaction_id = 0
-        self.debug = 1
+        self.debug = debug
         # Data Related Variables
         self.data = {}
         self.data_idx = []
@@ -44,7 +44,6 @@ class Network:
             return
 
         try:
-            logger.info("Connecting to %s" % self._server_address)
             if self._server_address.startswith("/"):
                 self.socket = socket.socket(socket.AF_UNIX, socket.SOCK_STREAM)
                 self.socket.connect(self._server_address)
@@ -55,9 +54,10 @@ class Network:
                 self.socket.connect((host, int(port)))
             self.socket.settimeout(self._timeout)
             self._connected = True
+            logger.info("Connected to %s" % self._server_address)
         except socket.error:
-            logger.error("Failed to connect to %s" % self._server_address)
             self._connected = False
+            logger.error("Failed to connect to %s" % self._server_address)
 
     def disconnect(self):
         if not self._connected:

vpp-snmp-agent.py

@@ -96,9 +96,14 @@ class MyAgent(agentx.Agent):
 
     def update(self):
         try:
+            self.vpp.connect()
+            r = self.vpp.vpp.api.control_ping()
+            self.logger.debug(f"VPP API: {r}")
             self.vppstat.connect()
-        except:
+        except Exception as e:
-            self.logger.error("Could not connect to VPPStats segment")
+            self.logger.error(f"VPP API: {e}, retrying")
+            self.vppstat.disconnect()
+            self.vpp.disconnect()
             return False
 
         ds = agentx.DataSet()
@@ -108,7 +113,6 @@ class MyAgent(agentx.Agent):
         num_ifaces = len(ifaces)
         num_vppstat = len(self.vppstat["/if/names"])
         num_lcp = len(lcp)
-        self.logger.debug("LCP: %s" % (lcp))
         self.logger.debug(
             "Retrieved Interfaces: vppapi=%d vppstat=%d lcp=%d"
             % (num_ifaces, num_vppstat, num_lcp)
@@ -384,10 +388,16 @@ def main():
     parser.add_argument(
         "-d", dest="debug", action="store_true", help="""Enable debug, default False"""
     )
+    parser.add_argument(
+        "-dd",
+        dest="debug_agent",
+        action="store_true",
+        help="""Enable agentx debug, default False""",
+    )
 
     args = parser.parse_args()
     if args.debug:
-        print("Arguments:", args)
+        print(f"Arguments: {args}")
 
     agentx.setup_logging(debug=args.debug)
 

vppapi.py

@@ -3,7 +3,7 @@ The functions in this file interact with the VPP API to retrieve certain
 interface metadata.
 """
 
-from vpp_papi import VPPApiClient
+from vpp_papi import VPPApiClient, VPPApiJSONFiles
 import os
 import fnmatch
 import logging
@@ -25,24 +25,36 @@ class VPPApi:
         self.connected = False
         self.clientname = clientname
         self.vpp = None
+        self.iface_dict = None
+        self.lcp_dict = None
+
+    def _sw_interface_event(self, event):
+        # NOTE(pim): this callback runs in a background thread, so we just clear the
+        # cached interfaces and LCPs here, subsequent call to get_ifaces() or get_lcp()
+        # will refresh them in the main thread.
+        logger.info(f"Clearing iface and LCP cache due to interface event")
+        self.iface_dict = None
+        self.lcp_dict = None
+
+    def _event_callback(self, msg_type_name, msg_type):
+        logger.debug(f"Received callback: {msg_type_name} => {msg_type}")
+        if msg_type_name == "sw_interface_event":
+            self._sw_interface_event(msg_type)
+        else:
+            logger.warning(f"Ignoring unkonwn event: {msg_type_name} => {msg_type}")
 
     def connect(self):
         if self.connected:
             return True
 
-        vpp_json_dir = "/usr/share/vpp/api/"
+        vpp_json_dir = VPPApiJSONFiles.find_api_dir([])
-
+        vpp_jsonfiles = VPPApiJSONFiles.find_api_files(api_dir=vpp_json_dir)
-        # construct a list of all the json api files
+        if not vpp_jsonfiles:
-        jsonfiles = []
-        for root, dirnames, filenames in os.walk(vpp_json_dir):
-            for filename in fnmatch.filter(filenames, "*.api.json"):
-                jsonfiles.append(os.path.join(root, filename))
-
-        if not jsonfiles:
             logger.error("no json api files found")
             return False
 
-        self.vpp = VPPApiClient(apifiles=jsonfiles, server_address=self.address)
+        self.vpp = VPPApiClient(apifiles=vpp_jsonfiles, server_address=self.address)
+        self.vpp.register_event_callback(self._event_callback)
         try:
             logger.info("Connecting to VPP")
             self.vpp.connect(self.clientname)
@@ -52,6 +64,13 @@ class VPPApi:
         v = self.vpp.api.show_version()
         logger.info("VPP version is %s" % v.version)
 
+        logger.info("Enabling VPP API interface events")
+        r = self.vpp.api.want_interface_events(enable_disable=True)
+        if r.retval != 0:
+            logger.error("Could not enable VPP API interface events, disconnecting")
+            self.disconnect()
+            return False
+
         self.connected = True
         return True
 
@@ -59,42 +78,58 @@ class VPPApi:
         if not self.connected:
             return True
         self.vpp.disconnect()
+        self.iface_dict = None
+        self.lcp_dict = None
         self.connected = False
         return True
 
     def get_ifaces(self):
         ret = {}
-        if not self.connected:
+        if not self.connected and not self.connect():
+            logger.warning("Can't connect to VPP API")
             return ret
 
+        if type(self.iface_dict) is dict:
+            logger.debug("Returning cached interfaces")
+            return self.iface_dict
+
+        ret = {}
         try:
+            logger.info("Requesting interfaces from VPP API")
             iface_list = self.vpp.api.sw_interface_dump()
         except Exception as e:
-            logger.error("VPP communication error, disconnecting", e)
+            logger.error("VPP API communication error, disconnecting", e)
-            self.vpp.disconnect()
+            self.disconnect()
-            self.connected = False
             return ret
 
         if not iface_list:
-            logger.error("Can't get interface list")
+            logger.error("Can't get interface list, disconnecting")
+            self.disconnect()
             return ret
 
         for iface in iface_list:
             ret[iface.interface_name] = iface
 
-        return ret
+        self.iface_dict = ret
+        logger.debug(f"Caching interfaces: {ret}")
+        return self.iface_dict
 
     def get_lcp(self):
         ret = {}
-        if not self.connected:
+        if not self.connected and not self.connect():
+            logger.warning("Can't connect to VPP API")
             return ret
 
+        if type(self.lcp_dict) is dict:
+            logger.debug("Returning cached LCPs")
+            return self.lcp_dict
+
         try:
+            logger.info("Requesting LCPs from VPP API")
             lcp_list = self.vpp.api.lcp_itf_pair_get()
         except Exception as e:
             logger.error("VPP communication error, disconnecting", e)
-            self.vpp.disconnect()
+            self.disconnect()
-            self.connected = False
             return ret
 
         if not lcp_list:
@@ -103,4 +138,7 @@ class VPPApi:
 
         for lcp in lcp_list[1]:
             ret[lcp.host_if_name] = lcp
-        return ret
+
+        self.lcp_dict = ret
+        logger.debug(f"Caching LCPs: {ret}")
+        return self.lcp_dict