Merge remote-tracking branch 'origin/dev' into feature/discord-bot

.github/workflows/bandit-action.yaml

@@ -5,13 +5,15 @@ on:
     branches:
       - main
       - dev
+      - feature/**
+      - feat/**
 
 jobs:
   test:
     runs-on: ubuntu-latest
     steps:
       - name: Bandit Test
-        uses: mdegis/bandit-action@v1
+        uses: mdegis/bandit-action@v1.0.1
         with:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
           path: "."

.github/workflows/consumer-docker.yaml

@@ -0,0 +1,42 @@
+name: Consumer CI
+
+on:
+    push:
+        branches: [main, dev, feature/consumer]
+    pull_request:
+        branches: [main, dev]
+
+env:
+  IMAGE_NAME: tsmc-hackathon-2024-consumer 
+  DOCKER_FILE_PATH: ./consumer # '.' for root directory
+
+jobs:
+    docker:
+        runs-on: ubuntu-latest
+        steps:
+            - name: Checkout
+              uses: actions/checkout@v4
+            - name: Set up QEMU
+              uses: docker/setup-qemu-action@v3
+            - name: Set up Docker Buildx
+              uses: docker/setup-buildx-action@v3
+            - name: Set up containerd
+              uses: crazy-max/ghaction-setup-containerd@v3
+            - name: Login to DockerHub
+              uses: docker/login-action@v3
+              with:
+                username: ${{ secrets.DOCKER_USERNAME }}
+                password: ${{ secrets.DOCKER_PASSWORD }}
+            - name: Build Docker image
+              uses: docker/build-push-action@v4
+              with:
+                context: ${{ env.DOCKER_FILE_PATH }}
+                platforms: linux/amd64,linux/arm64
+                tags: docker.io/${{ secrets.DOCKER_USERNAME }}/${{ env.IMAGE_NAME }}:latest
+                outputs: type=oci,dest=/tmp/image.tar
+            - name: Import image in containerd
+              run: |
+                sudo ctr i import --base-name docker.io/${{ secrets.DOCKER_USERNAME }}/${{ env.IMAGE_NAME }}:latest --digests --all-platforms /tmp/image.tar
+            - name: Push image to containerd
+              run: |
+                sudo ctr i push --user "${{ secrets.DOCKER_USERNAME }}:${{ secrets.DOCKER_PASSWORD }}" docker.io/${{ secrets.DOCKER_USERNAME }}/${{ env.IMAGE_NAME }}:latest

.gitignore

@@ -158,3 +158,5 @@ cython_debug/
 #  and can be added to the global gitignore or merged into this file.  For a more nuclear
 #  option (not recommended) you can uncomment the following to ignore the entire idea folder.
 #.idea/
+
+.env/*

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