Salmanap/fix network agent demo (#153)

* staging my changes to re-based from main

* adding debug statements to rust

* merged with main

* ready to push network agent

* removed the incomplete sql example

---------

Co-authored-by: Salman Paracha <salmanparacha@MacBook-Pro-261.local>
This commit is contained in:
Salman Paracha 2024-10-08 22:19:20 -07:00 committed by GitHub
parent 6acfea7787
commit b63a01fe82
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
41 changed files with 252 additions and 1987 deletions

View file

@ -0,0 +1,19 @@
FROM python:3.10 AS base
FROM base AS builder
WORKDIR /src
COPY requirements.txt /src/
RUN pip install --prefix=/runtime --force-reinstall -r requirements.txt
COPY . /src
FROM python:3.10-slim AS output
COPY --from=builder /runtime /usr/local
COPY . /app
WORKDIR /app
CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "80", "--log-level", "info"]

View file

View file

@ -0,0 +1,71 @@
version: v0.1
listener:
address: 127.0.0.1
port: 8080 #If you configure port 443, you'll need to update the listener with tls_certificates
message_format: huggingface
# Centralized way to manage LLMs, manage keys, retry logic, failover and limits in a central way
llm_providers:
- name: OpenAI
provider: openai
access_key: OPENAI_API_KEY
model: gpt-4o
default: true
# default system prompt used by all prompt targets
system_prompt: |
You are a network assistant that just offers facts; not advice on manufacturers or purchasing decisions.
prompt_targets:
- name: reboot_devices
description: Reboot specific devices or device groups
endpoint:
name: app_server
path: /agent/device_reboot
parameters:
- name: device_ids
type: list
description: A list of device identifiers (IDs) to reboot.
required: true
- name: time_range
type: int
description: Optional time range in days for reboot operations. Defaults to 7.
- name: network_qa
endpoint:
name: app_server
path: /agent/network_summary
description: Handle general Q/A related to networking.
default: true
- name: device_summary
description: Retrieve statistics for specific devices within a time range
endpoint:
name: app_server
path: /agent/device_summary
parameters:
- name: device_ids
type: list
description: A list of device identifiers (IDs) to retrieve statistics for.
required: true # device_ids are required to get device statistics
- name: time_range
type: int
description: Time range in days for which to gather device statistics. Defaults to 7.
default: "7"
# Arch creates a round-robin load balancing between different endpoints, managed via the cluster subsystem.
endpoints:
app_server:
# value could be ip address or a hostname with port
# this could also be a list of endpoints for load balancing
# for example endpoint: [ ip1:port, ip2:port ]
endpoint: host.docker.internal:18083
# max time to wait for a connection to be established
connect_timeout: 0.005s
ratelimits:
- model: gpt-4
selector:
key: selector-key
value: selector-value
limit:
tokens: 1
unit: minute

View file

@ -0,0 +1,21 @@
services:
api_server:
build:
context: .
dockerfile: Dockerfile
ports:
- "18083:80"
healthcheck:
test: ["CMD", "curl" ,"http://localhost:80/healthz"]
interval: 5s
retries: 20
chatbot_ui:
build:
context: ../../chatbot_ui
dockerfile: Dockerfile
ports:
- "18080:8080"
environment:
- OPENAI_API_KEY=${OPENAI_API_KEY:?error}
- CHAT_COMPLETION_ENDPOINT=http://host.docker.internal:10000/v1

View file

@ -0,0 +1,12 @@
apiVersion: 1
providers:
- name: "Dashboard provider"
orgId: 1
type: file
disableDeletion: false
updateIntervalSeconds: 10
allowUiUpdates: false
options:
path: /var/lib/grafana/dashboards
foldersFromFilesStructure: true

View file

@ -0,0 +1,355 @@
{
"annotations": {
"list": [
{
"builtIn": 1,
"datasource": {
"type": "grafana",
"uid": "-- Grafana --"
},
"enable": true,
"hide": true,
"iconColor": "rgba(0, 211, 255, 1)",
"name": "Annotations & Alerts",
"type": "dashboard"
}
]
},
"editable": true,
"fiscalYearStartMonth": 0,
"graphTooltip": 1,
"links": [],
"panels": [
{
"datasource": {
"type": "prometheus",
"uid": "PBFA97CFB590B2093"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "red",
"value": 80
}
]
}
},
"overrides": []
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 0
},
"id": 2,
"options": {
"legend": {
"calcs": [],
"displayMode": "list",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"mode": "single",
"sort": "none"
}
},
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "PBFA97CFB590B2093"
},
"disableTextWrap": false,
"editorMode": "code",
"expr": "avg(rate(envoy_cluster_internal_upstream_rq_time_sum[1m]) / rate(envoy_cluster_internal_upstream_rq_time_count[1m])) by (envoy_cluster_name)",
"fullMetaSearch": false,
"hide": false,
"includeNullMetadata": true,
"instant": false,
"legendFormat": "__auto",
"range": true,
"refId": "A",
"useBackend": false
}
],
"title": "request latency - internal (ms)",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "PBFA97CFB590B2093"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "red",
"value": 80
}
]
}
},
"overrides": []
},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 0
},
"id": 1,
"options": {
"legend": {
"calcs": [],
"displayMode": "list",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"mode": "single",
"sort": "none"
}
},
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "PBFA97CFB590B2093"
},
"disableTextWrap": false,
"editorMode": "code",
"expr": "avg(rate(envoy_cluster_external_upstream_rq_time_sum[1m]) / rate(envoy_cluster_external_upstream_rq_time_count[1m])) by (envoy_cluster_name)",
"fullMetaSearch": false,
"hide": false,
"includeNullMetadata": true,
"instant": false,
"legendFormat": "__auto",
"range": true,
"refId": "A",
"useBackend": false
}
],
"title": "request latency - external (ms)",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "PBFA97CFB590B2093"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"drawStyle": "line",
"fillOpacity": 0,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "linear",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "auto",
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": null
},
{
"color": "red",
"value": 80
}
]
}
},
"overrides": []
},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 8
},
"id": 3,
"options": {
"legend": {
"calcs": [],
"displayMode": "list",
"placement": "bottom",
"showLegend": true
},
"tooltip": {
"mode": "single",
"sort": "none"
}
},
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "PBFA97CFB590B2093"
},
"disableTextWrap": false,
"editorMode": "code",
"expr": "avg(rate(envoy_cluster_internal_upstream_rq_completed[1m])) by (envoy_cluster_name)",
"fullMetaSearch": false,
"includeNullMetadata": true,
"instant": false,
"legendFormat": "__auto",
"range": true,
"refId": "A",
"useBackend": false
},
{
"datasource": {
"type": "prometheus",
"uid": "PBFA97CFB590B2093"
},
"disableTextWrap": false,
"editorMode": "code",
"expr": "avg(rate(envoy_cluster_external_upstream_rq_completed[1m])) by (envoy_cluster_name)",
"fullMetaSearch": false,
"hide": false,
"includeNullMetadata": true,
"instant": false,
"legendFormat": "__auto",
"range": true,
"refId": "B",
"useBackend": false
}
],
"title": "Upstream request count",
"type": "timeseries"
}
],
"schemaVersion": 39,
"tags": [],
"templating": {
"list": []
},
"time": {
"from": "now-15m",
"to": "now"
},
"timepicker": {},
"timezone": "browser",
"title": "Intelligent Gateway Overview",
"uid": "adt6uhx5lk8aob",
"version": 3,
"weekStart": ""
}

View file

@ -0,0 +1,9 @@
apiVersion: 1
datasources:
- name: Prometheus
type: prometheus
url: http://prometheus:9090
isDefault: true
access: proxy
editable: true

104
demos/network_agent/main.py Normal file
View file

@ -0,0 +1,104 @@
from fastapi import FastAPI, HTTPException
from pydantic import BaseModel, Field
from typing import List, Optional
app = FastAPI()
# Define the request model
class DeviceSummaryRequest(BaseModel):
device_ids: List[int]
time_range: Optional[int] = Field(default=7, description="Time range in days, defaults to 7")
# Define the response model
class DeviceStatistics(BaseModel):
device_id: int
time_range: str
data: str
class DeviceSummaryResponse(BaseModel):
statistics: List[DeviceStatistics]
# Request model for device reboot
class DeviceRebootRequest(BaseModel):
device_ids: List[int]
# Response model for the device reboot
class CoverageResponse(BaseModel):
status: str
summary: dict
@app.post("/agent/device_reboot", response_model=CoverageResponse)
def reboot_network_device(request_data: DeviceRebootRequest):
"""
Endpoint to reboot network devices based on device IDs and an optional time range.
"""
# Access data from the Pydantic model
device_ids = request_data.device_ids
# Validate 'device_ids' (This is already validated by Pydantic, but additional logic can be added if needed)
if not device_ids:
raise HTTPException(status_code=400, detail="'device_ids' parameter is required")
# Simulate reboot operation and return the response
statistics = []
for device_id in device_ids:
# Placeholder for actual data retrieval or device reboot logic
stats = {
"data": f"Device {device_id} has been successfully rebooted."
}
statistics.append(stats)
# Return the response with a summary
return CoverageResponse(status="success", summary={"device_ids": device_ids})
# Post method for device summary
@app.post("/agent/device_summary", response_model=DeviceSummaryResponse)
def get_device_summary(request: DeviceSummaryRequest):
"""
Endpoint to retrieve device statistics based on device IDs and an optional time range.
"""
# Extract 'device_ids' and 'time_range' from the request
device_ids = request.device_ids
time_range = request.time_range
# Simulate retrieving statistics for the given device IDs and time range
statistics = []
minutes = 1
for device_id in device_ids:
stats = {
"device_id": device_id,
"time_range": f"Last {time_range} days",
"data": f"Device {device_id} over the last {time_range} days experienced {minutes} minutes of downtime.",
}
minutes += 1
statistics.append(DeviceStatistics(**stats))
return DeviceSummaryResponse(statistics=statistics)
@app.post("/agent/network_summary")
async def policy_qa():
"""
This method handles Q/A related to general issues in networks.
It forwards the conversation to the OpenAI client via a local proxy and returns the response.
"""
return {
"choices": [
{
"message": {
"role": "assistant",
"content": "I am a helpful networking agent, and I can help you get status for network devices or reboot them"
},
"finish_reason": "completed",
"index": 0
}
],
"model": "network_agent",
"usage": {
"completion_tokens": 0
}
}
if __name__ == "__main__":
app.run(debug=True)

View file

@ -0,0 +1,4 @@
fastapi
uvicorn
pydantic
typing

View file

@ -0,0 +1,247 @@
import pandas as pd
import random
from datetime import datetime, timedelta, timezone
import re
import logging
from dateparser import parse
import sqlite3
logging.basicConfig(
level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s"
)
logger = logging.getLogger(__name__)
def load_sql():
# Example Usage
conn = sqlite3.connect(":memory:")
# create and load the devices table
device_data = generate_device_data(conn)
# create and load the interface_stats table
generate_interface_stats_data(conn, device_data)
# create and load the flow table
generate_flow_data(conn, device_data)
return conn
# Function to convert natural language time expressions to "X {time} ago" format
def convert_to_ago_format(expression):
# Define patterns for different time units
time_units = {
r"seconds": "seconds",
r"minutes": "minutes",
r"mins": "mins",
r"hrs": "hrs",
r"hours": "hours",
r"hour": "hour",
r"hr": "hour",
r"days": "days",
r"day": "day",
r"weeks": "weeks",
r"week": "week",
r"months": "months",
r"month": "month",
r"years": "years",
r"yrs": "years",
r"year": "year",
r"yr": "year",
}
# Iterate over each time unit and create regex for each phrase format
for pattern, unit in time_units.items():
# Handle "for the past X {unit}"
match = re.search(rf"(\d+) {pattern}", expression)
if match:
quantity = match.group(1)
return f"{quantity} {unit} ago"
# If the format is not recognized, return None or raise an error
return None
# Function to generate random MAC addresses
def random_mac():
return "AA:BB:CC:DD:EE:" + ":".join(
[f"{random.randint(0, 255):02X}" for _ in range(2)]
)
# Function to generate random IP addresses
def random_ip():
return f"{random.randint(1, 255)}.{random.randint(1, 255)}.{random.randint(1, 255)}.{random.randint(1, 255)}"
# Generate synthetic data for the device table
def generate_device_data(
conn,
n=1000,
):
device_data = {
"switchip": [random_ip() for _ in range(n)],
"hwsku": [f"HW{i+1}" for i in range(n)],
"hostname": [f"switch{i+1}" for i in range(n)],
"osversion": [f"v{i+1}" for i in range(n)],
"layer": ["L2" if i % 2 == 0 else "L3" for i in range(n)],
"region": [random.choice(["US", "EU", "ASIA"]) for _ in range(n)],
"uptime": [
f"{random.randint(0, 10)} days {random.randint(0, 23)}:{random.randint(0, 59)}:{random.randint(0, 59)}"
for _ in range(n)
],
"device_mac_address": [random_mac() for _ in range(n)],
}
df = pd.DataFrame(device_data)
df.to_sql("device", conn, index=False)
return df
# Generate synthetic data for the interfacestats table
def generate_interface_stats_data(conn, device_df, n=1000):
interface_stats_data = []
for _ in range(n):
device_mac = random.choice(device_df["device_mac_address"])
ifname = random.choice(["eth0", "eth1", "eth2", "eth3"])
time = datetime.now(timezone.utc) - timedelta(
minutes=random.randint(0, 1440 * 5)
) # random timestamps in the past 5 day
in_discards = random.randint(0, 1000)
in_errors = random.randint(0, 500)
out_discards = random.randint(0, 800)
out_errors = random.randint(0, 400)
in_octets = random.randint(1000, 100000)
out_octets = random.randint(1000, 100000)
interface_stats_data.append(
{
"device_mac_address": device_mac,
"ifname": ifname,
"time": time,
"in_discards": in_discards,
"in_errors": in_errors,
"out_discards": out_discards,
"out_errors": out_errors,
"in_octets": in_octets,
"out_octets": out_octets,
}
)
df = pd.DataFrame(interface_stats_data)
df.to_sql("interfacestats", conn, index=False)
return
# Generate synthetic data for the ts_flow table
def generate_flow_data(conn, device_df, n=1000):
flow_data = []
for _ in range(n):
sampler_address = random.choice(device_df["switchip"])
proto = random.choice(["TCP", "UDP"])
src_addr = random_ip()
dst_addr = random_ip()
src_port = random.randint(1024, 65535)
dst_port = random.randint(1024, 65535)
in_if = random.randint(1, 10)
out_if = random.randint(1, 10)
flow_start = int(
(datetime.now() - timedelta(days=random.randint(1, 30))).timestamp()
)
flow_end = int(
(datetime.now() - timedelta(days=random.randint(1, 30))).timestamp()
)
bytes_transferred = random.randint(1000, 100000)
packets = random.randint(1, 1000)
flow_time = datetime.now(timezone.utc) - timedelta(
minutes=random.randint(0, 1440 * 5)
) # random flow time
flow_data.append(
{
"sampler_address": sampler_address,
"proto": proto,
"src_addr": src_addr,
"dst_addr": dst_addr,
"src_port": src_port,
"dst_port": dst_port,
"in_if": in_if,
"out_if": out_if,
"flow_start": flow_start,
"flow_end": flow_end,
"bytes": bytes_transferred,
"packets": packets,
"time": flow_time,
}
)
df = pd.DataFrame(flow_data)
df.to_sql("ts_flow", conn, index=False)
return
def load_params(req):
# Step 1: Convert the from_time natural language string to a timestamp if provided
if req.from_time:
# Use `dateparser` to parse natural language timeframes
logger.info(f"{'* ' * 50}\n\nCaptured from time: {req.from_time}\n\n")
parsed_time = parse(req.from_time, settings={"RELATIVE_BASE": datetime.now()})
if not parsed_time:
conv_time = convert_to_ago_format(req.from_time)
if conv_time:
parsed_time = parse(
conv_time, settings={"RELATIVE_BASE": datetime.now()}
)
else:
return {
"error": "Invalid from_time format. Please provide a valid time description such as 'past 7 days' or 'since last month'."
}
logger.info(f"\n\nConverted from time: {parsed_time}\n\n{'* ' * 50}\n\n")
from_time = parsed_time
logger.info(f"Using parsed from_time: {from_time}")
else:
# If no from_time is provided, use a default value (e.g., the past 7 days)
from_time = datetime.now() - timedelta(days=7)
logger.info(f"Using default from_time: {from_time}")
# Step 2: Build the dynamic SQL query based on the optional filters
filters = []
params = {"from_time": from_time}
if req.ifname:
filters.append("i.ifname = :ifname")
params["ifname"] = req.ifname
if req.region:
filters.append("d.region = :region")
params["region"] = req.region
if req.min_in_errors is not None:
filters.append("i.in_errors >= :min_in_errors")
params["min_in_errors"] = req.min_in_errors
if req.max_in_errors is not None:
filters.append("i.in_errors <= :max_in_errors")
params["max_in_errors"] = req.max_in_errors
if req.min_out_errors is not None:
filters.append("i.out_errors >= :min_out_errors")
params["min_out_errors"] = req.min_out_errors
if req.max_out_errors is not None:
filters.append("i.out_errors <= :max_out_errors")
params["max_out_errors"] = req.max_out_errors
if req.min_in_discards is not None:
filters.append("i.in_discards >= :min_in_discards")
params["min_in_discards"] = req.min_in_discards
if req.max_in_discards is not None:
filters.append("i.in_discards <= :max_in_discards")
params["max_in_discards"] = req.max_in_discards
if req.min_out_discards is not None:
filters.append("i.out_discards >= :min_out_discards")
params["min_out_discards"] = req.min_out_discards
if req.max_out_discards is not None:
filters.append("i.out_discards <= :max_out_discards")
params["max_out_discards"] = req.max_out_discards
return params, filters