Merge Discord, ai, monitor and provide a sample a runner for demo

Discord-Bot/bot.py

@@ -1,6 +1,15 @@
 import discord
-# 連結 Error_Notify
-# 連接 Receive_Mess
+import random
+import asyncio
+import os 
+from message import send_embedded_warning, send_embedded_error,send_embedded_info
+from feedback import get_active_threads, process_feedback
+from dotenv import load_dotenv
+load_dotenv()
+DISCORD_BOT_TOKEN = os.getenv('DISCORD_BOT_TOKEN')
+
+last_warning = None
+active_threads = []
 
 # 權限設置
 intents = discord.Intents.default()
@@ -9,13 +18,57 @@
 
 client = discord.Client(intents = intents)
 
+async def update_active_threads():
+    global active_threads
+    active_threads = await get_active_threads()
+
 @client.event
 async def on_ready():
-    slash = await client.tree.sync()
-    print("TSMC System Bot is Online!")
     print(f'Logged in as {client.user}')
+    test_channel_id = 1199372364870340810  
+    channel = client.get_channel(test_channel_id)
+    client.loop.create_task(update_active_threads())
+    if channel:
+        await send_embedded_warning(channel)
+        # await asyncio.sleep(5)
+        # await send_embedded_error(channel)
+        # await asyncio.sleep(5)
+        # await send_embedded_info(channel)
+
+# 監聽討論串訊息
+@client.event
+async def on_message(message):
+    if message.author == client.user:
+        return
 
+    for thread in active_threads:
+        if message.channel.id == thread.id:
+            await process_feedback(message, thread)
+
+
+# 測試 function
+# async def pull_warning():
+#     num = random.randint(0,100)
+#     return f"編號{num}這是一個新的警告！"
+
+# async def warning_task():
+#     global last_warning
+#     while True:
+#         # 呼叫 pull_warning 函數獲取最新警告
+#         new_warning = await pull_warning()
+#         if new_warning != last_warning:
+#             # 如果新警告與上一次警告不同，則向 Discord 頻道發送新警告
+#             last_warning = new_warning
+#             channel = client.get_channel(1199372364870340810)  
+#             await channel.send(new_warning)
+#         await asyncio.sleep(5)  
+
 
+# @client.event
+# async def on_ready():
+#     print(f'{client.user} has connected to Discord!')
+#     print("TSMC System Bot is Online!")
+#     client.loop.create_task(warning_task())
 
 
-client.run("MTE5OTI3MTYxMzM1OTczOTAxMA.GJN31L.rNSDr17bJ4cMPOsKeLd4jFdS45wZqVGcrDmo6k")
+client.run(DISCORD_BOT_TOKEN)

Discord-Bot/feedback.py

@@ -0,0 +1,29 @@
+import discord
+
+active_threads = []
+
+async def get_active_threads():
+    return active_threads  
+
+async def send_feedback(content):
+    # 看 Jerry 那邊要怎麼收
+    pass
+
+async def create_thread(message):
+    # 創建討論串
+    thread = await message.create_thread(name="Feedback Discussion")
+    # 在討論串中發送一則 Welcome 訊息
+    await thread.send("Send me message if you have any suggestion!")
+    # 將討論串加入 active_threads 之中
+    active_threads.append(thread)
+    print(active_threads)
+
+
+async def process_feedback(message, thread):
+    await thread.send(f"This is your feedback\n {message.content}\n Thanks for your feedback, Jerry will take care of it!")
+    print("成功發送「確認收到feedback訊息」")
+    # await send_feedback(message)
+
+
+
+

Discord-Bot/message.py

@@ -0,0 +1,84 @@
+import discord
+
+
+async def send_embedded_warning(channel):
+    # 創建一個 Embed 對象
+    embed = discord.Embed(
+        title = "WARNING",  
+        description = "We're out of CPU !!!", 
+        color = 0xf31cc8
+    )
+
+    # 添加欄位
+    embed.add_field(
+        name = "Suggestion1: Auto Scale \n",
+        value = "We have to auto-scale the CPU to make the system keep working.",
+        inline = False
+    )
+    embed.add_field(
+        name = "Suggestion2: Give up the system \n",
+        value = "There's nothing we can do bruh",
+        inline = False
+    )
+
+    # 設置腳註
+    embed.set_footer(text = "Warning Message by your system buddy")
+
+    message = await channel.send(embed = embed)
+
+    # 創建討論串
+    from feedback import create_thread
+    await create_thread(message)
+
+
+
+async def send_embedded_error(channel):
+    # 創建一個 Embed 對象
+    embed = discord.Embed(
+        title = "ERROR",  
+        description = "We encounter a Network Error!!!",  
+        color = discord.Color.red()  
+    )
+
+    # 添加欄位
+    embed.add_field(
+        name = "Suggestion1: Go get 張鴈光's help \n",
+        value = "He's the man in charge of CSIE wifi, but the wifi is always suck : (",
+        inline = False
+    )
+    embed.add_field(
+        name = "Suggestion2: Give up the system \n",
+        value = "There's nothing we can do bruh",
+        inline = False
+    )
+
+    # 設置腳註
+    embed.set_footer(text = "Error Message by your system buddy")
+
+    await channel.send(embed = embed)
+
+async def send_embedded_info(channel):
+    # 創建一個 Embed 對象
+    embed = discord.Embed(
+        title = "Info",  
+        description = "Hey! Here's some info",  
+        color = 0x25f533  
+    )
+
+    # 添加欄位
+    embed.add_field(
+        name = "Info1: We're going to win the competition! \n",
+        value = "We're the best!",
+        inline = False
+    )
+    embed.add_field(
+        name = "Info2: we're bringing ipad back! \n",
+        value = "We're the best!!",
+        inline = False
+    )
+
+    # 設置腳註
+    embed.set_footer(text = "Info Message by your system buddy")
+
+    await channel.send(embed = embed)
+

ai/analyze.py

@@ -0,0 +1,154 @@
+import pandas as pd
+from langchain.output_parsers import ResponseSchema, StructuredOutputParser
+from langchain.prompts import PromptTemplate
+from langchain_google_vertexai import VertexAI
+
+
+def analyze_cpu_usage(metric_df) -> str:
+    feedback = ""
+    label = "Container CPU Utilization (%)"
+    for i in range(len(metric_df) - 1):
+        curr_entry = metric_df.iloc[i][label]
+        next_entry = metric_df.iloc[i + 1][label]
+        if (
+            not pd.isna(curr_entry)
+            and not pd.isna(next_entry)
+            and curr_entry > 60.0
+            and next_entry > 60.0
+        ):
+            feedback += f"\
+- ERROR: Container CPU Utilization (%) is above 60% for two minutes, at {metric_df.iloc[i]['Time']} and {metric_df.iloc[i + 1]['Time']}\n"
+
+    if feedback == "":
+        feedback = f"- INFO: Container CPU Utilization (%) is below 60% over the last {len(metric_df)} minutes.\n"
+
+    return feedback
+
+
+def analyze_mem_usage(metric_df) -> str:
+    feedback = ""
+    label = "Container Memory Utilization (%)"
+    for i in range(len(metric_df) - 1):
+        curr_entry = metric_df.iloc[i][label]
+        next_entry = metric_df.iloc[i + 1][label]
+        if (
+            not pd.isna(curr_entry)
+            and not pd.isna(next_entry)
+            and curr_entry > 60.0
+            and next_entry > 60.0
+        ):
+            feedback += f"\
+- ERROR: Container Memory Utilization (%) is above 60% for two minutes, at {metric_df.iloc[i]['Time']} and {metric_df.iloc[i + 1]['Time']}\n"
+
+    if feedback == "":
+        feedback = f"- INFO: Container Memory Utilization (%) is below 60% over the last {len(metric_df)} minutes.\n"
+
+    return feedback
+
+
+def analyze_restart(metric_df) -> str:
+    feedback = ""
+    for _, row in metric_df.iterrows():
+        if not pd.isna(row["Container Startup Latency (ms)"]):
+            feedback += f"\
+- ERROR: Cloud run restarted at {row['Time']}, with Container Startup Latency (ms) of {row['Container Startup Latency (ms)']} ms\n"
+
+    if feedback == "":
+        feedback = f"- INFO: Cloud run did not restart over the last {len(metric_df)} minutes.\n"
+
+    return feedback
+
+
+def analyze_instance_count(metric_df) -> str:
+    feedback = ""
+    for _, row in metric_df.iterrows():
+        if pd.isna(row["Instance Count (active)"]):
+            continue
+
+        total_instance_count = (
+            row["Instance Count (active)"] + row["Instance Count (idle)"]
+        )
+        if total_instance_count > 2:
+            feedback += f"\
+- ERROR: Total instance count is above 2 at {row['Time']}, with Instance Count (active) of {int(row['Instance Count (active)'])} and Instance Count (idle) of {int(row['Instance Count (idle)'])}\n"
+
+    if feedback == "":
+        feedback = f"- INFO: Total instance count is less than or equal to 2 over the last {len(metric_df)} minutes.\n"
+
+    return feedback
+
+
+def analyze_by_rule(metric_df: pd.DataFrame) -> str:
+    feedback = ""
+    feedback += analyze_cpu_usage(metric_df)
+    feedback += analyze_mem_usage(metric_df)
+    feedback += analyze_restart(metric_df)
+    feedback += analyze_instance_count(metric_df)
+
+    return feedback
+
+
+def analyze_by_llm(metric_df: pd.DataFrame) -> dict:
+    # Analysis feedback by heuristic rules
+    heuristic_feedback = analyze_by_rule(metric_df)
+
+    # Define response schema
+    severity_schema = ResponseSchema(
+        name="severity",
+        description='Severity level of the analysis feedback. \
+Use "ERROR" if the analysis detects errors, "WARNING" for potential issues, or "INFO" if no problems are identified.',
+    )
+    message_schema = ResponseSchema(
+        name="message",
+        description="In-depth analysis feedback based on provided metrics(The description can span multiple lines, use '\\n' to separate lines.)",
+    )
+    response_schema = [severity_schema, message_schema]
+    output_parser = StructuredOutputParser.from_response_schemas(response_schema)
+    format_instruction = output_parser.get_format_instructions()
+
+    # Define the model and prompt template
+    llm = VertexAI(
+        model_name="text-bison@001",
+        temperature=0,
+        max_output_tokens=512,
+        top_p=0.8,
+        top_k=40,
+    )
+    prompt_template = PromptTemplate.from_template(
+        """\
+The following text contains metric data for a Google Cloud Run application. \
+This data is presented in CSV format and encompasses the most recent {time_span} minutes:
+{metric_data}
+
+The following text is a heuristic analysis feedback of the metric data:
+{heuristic_feedback}
+
+The heuristic analysis feedback is based on the following rules:
+- CPU limit > 60% (lasts 2 minutes)
+- Memory limit > 60% (lasts 2 minutes)
+- Cloud run re-start
+- Instance count > 2
+- Fail response (4xx, 5xx)
+
+Based on the provided metrics, an in-depth \
+analysis is required to evaluate the cloud resource status and the operational health of the system. The analysis \
+should identify and report any errors, anticipate potential problems, and propose appropriate remediation strategies.
+
+{format_instruction}
+"""
+    )
+
+    # Invoke the model
+    chain = prompt_template | llm
+    feedback = chain.invoke(
+        {
+            "time_span": len(metric_df),
+            "metric_data": metric_df.to_string(),
+            "heuristic_feedback": heuristic_feedback,
+            "format_instruction": format_instruction,
+        }
+    )
+
+    # Parse the feedback to a dictionary
+    feedback_dict = output_parser.parse(feedback)
+    return feedback_dict