如何通过Python代码在Mininet中定期记录所有流表条目？POX类查询

To programmatically record flow table entries in POX (instead of relying on dpctl dump-flows), you'll want to work with POX's OpenFlow stats message classes and the core openflow module. Here's the breakdown:

Key Classes & Components to Focus On #

• ofp_stats_request: This is the OpenFlow message you send to switches to request flow table statistics. Configure it with type=OFPST_FLOW to target flow entries specifically.
• ofp_flow_stats_reply: The response message from switches that contains the actual flow table data. Each entry in the reply is an ofp_flow_stats object, holding details like match criteria, actions, byte/packet counts, and flow duration.
• openflow.Connection: Represents the active connection to a Mininet switch—you’ll use this to send your stats requests and receive replies.
• POX's built-in flow_stats component: A pre-written module that handles basic flow stats collection. You can use it directly as a starting point, or dissect its code to build custom recording logic tailored to your needs.

How to Implement Flow Recording #

Here’s a simplified, actionable workflow to get you started:

1. Listen for switch connections: Use POX’s event system to detect when a switch connects to the controller (via the ConnectionUp event).
2. Schedule periodic flow stats requests: For each connected switch, set up a recurring timer to send ofp_stats_request messages—this ensures you capture frequent snapshots of the flow table.
3. Process and store the reply: When you receive an ofp_flow_stats_reply, extract each flow entry and write it to a log file, database, or any storage solution you prefer.

Quick Example Snippet #

from pox.core import core
import pox.openflow.libopenflow_01 as of
from pox.lib.recoco import Timer
from datetime import datetime

log = core.getLogger()

class FlowRecorder(object):
    def __init__(self):
        core.openflow.addListeners(self)
        self.switch_timers = {}  # Track timers for each connected switch

    def _handle_ConnectionUp(self, event):
        log.info("Switch %s connected. Starting flow recording...", event.dpid)
        # Send flow stats request every 3 seconds (adjust interval as needed)
        self.switch_timers[event.dpid] = Timer(3, self.request_flow_stats, args=[event.connection], recurring=True)

    def request_flow_stats(self, connection):
        # Build and send the flow stats request message
        stats_msg = of.ofp_stats_request()
        stats_msg.body = of.ofp_flow_stats_request()
        connection.send(stats_msg)

    def _handle_StatsReply(self, event):
        # Check if the reply contains flow stats
        if isinstance(event.stats, of.ofp_flow_stats_reply):
            timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
            # Append to a switch-specific log file
            with open(f"flow_logs/switch_{event.connection.dpid}.log", "a") as log_file:
                log_file.write(f"=== Flow Table Snapshot - {timestamp} ===\n")
                for flow in event.stats.entries:
                    log_file.write(f"Match: {flow.match} | Actions: {flow.actions} | Bytes: {flow.byte_count}\n")
                log_file.write("\n")

def launch():
    core.registerNew(FlowRecorder)
    log.info("Flow Recorder component initialized")

Notes #

• Adjust the timer interval (3 seconds in the example) based on how frequently you need to capture flow changes.
• The ofp_flow_stats object includes all the same details you see in dpctl dump-flows—check POX’s OpenFlow 1.0 docs for a full list of available fields (like priority, idle timeout, etc.).
• If you only want to record specific flows (e.g., those matching a certain IP or port), modify the ofp_flow_stats_request to include custom match criteria.

内容的提问来源于stack exchange，提问作者Sangeetha Sangeetha