Dashboard (#68)

* dashboard v0

* websocket logs

* logging

* update

* dockerfiles + docker-compose

* backend.default.svc.cluster.local:5001 (broken APIs)

* v0.5 dashboard

* dashboard 0.7

* dashboard v0.9

* DASHBOARD V1.0

* dashboard 1.01

* fixes v1

* fixes

* dashboard fixes minus socketio

* DASHBOARD V2

* dashboard logger

* small cleanup

* final?

* real final

* namespace route api fix?

* DASHBOARD FINAL

Dockerfile.backend

@@ -0,0 +1,57 @@
+# Use Python 3.11 slim as the base image
+FROM python:3.11-slim AS base
+
+# Set environment variables for Python behavior
+ENV PYTHONFAULTHANDLER=1 \
+    PYTHONHASHSEED=random \
+    PYTHONUNBUFFERED=1
+
+# Set the working directory inside the container
+WORKDIR /app
+
+# Builder stage: Install dependencies and build the backend package
+FROM base AS builder
+
+# Set environment variables for pip and Poetry
+ENV PIP_DEFAULT_TIMEOUT=100 \
+    PIP_DISABLE_PIP_VERSION_CHECK=1 \
+    PIP_NO_CACHE_DIR=1 \
+    POETRY_VERSION=1.3.1
+
+# Install Poetry
+RUN pip install "poetry==$POETRY_VERSION"
+
+# Copy the pyproject.toml and poetry.lock files
+COPY pyproject.toml poetry.lock ./
+
+# Copy the entire konduktor directory to the container
+COPY konduktor ./konduktor
+
+# List the contents of the konduktor directory to verify the copy
+RUN ls -la ./konduktor
+
+# Configure Poetry and install dependencies only for the dashboard group
+RUN poetry config virtualenvs.in-project true && \
+    poetry install --with dashboard --no-root
+
+# Final stage for production
+FROM base AS final
+
+# Set the working directory
+WORKDIR /app
+
+# Copy the virtual environment from the builder stage
+COPY --from=builder /app/.venv ./.venv
+
+# Copy the konduktor directory from the builder stage
+COPY --from=builder /app/konduktor ./konduktor
+
+# Copy the startup script
+COPY startup.sh /app/startup.sh
+RUN chmod +x /app/startup.sh
+
+# Expose the port the app runs on
+EXPOSE 5001
+
+# Set the startup command
+CMD ["/app/startup.sh"]

Dockerfile.frontend

@@ -0,0 +1,23 @@
+# Use the official Node.js 18 slim image
+FROM node:18-slim
+
+# Set the working directory
+WORKDIR /app
+
+# Copy package.json and package-lock.json from the /frontend folder
+COPY konduktor/dashboard/frontend/package*.json ./
+
+# Install dependencies
+RUN npm install
+
+# Copy the entire frontend source code
+COPY konduktor/dashboard/frontend/ .
+
+# Build the frontend for production
+RUN npm run build
+
+# Expose the frontend port
+EXPOSE 5173
+
+# Start the frontend app
+CMD ["npm", "run", "start"]

docs/source/admin/observability.rst

@@ -75,4 +75,37 @@ As well as (S)Xid errors by following :code:`dmesg` on each node. You can also p
 .. figure:: ../images/otel-loki.png
    :width: 120%
    :align: center
-   :alt: dashboard
+   :alt: dashboard
+
+Dashboard
+------------
+
+This is a user-friendly localhost dashboard to manage workloads within a cluster, all in one place. 
+
+Features include:
+
+- Grafana konduktor dashboard
+- Loki logs (search + filtering by namespace)
+- Table to view, delete, and modify priority of workloads in queue
+
+To open the dashboard, run this inside the root konduktor directory:
+
+.. code-block:: console
+
+    $ ./start_dashboard.sh
+
+If running into a permission error, try this instead:
+
+.. code-block:: console
+
+    $ chmod +x start_dashboard.sh && ./start_dashboard.sh
+
+.. figure:: ../images/dashboard-logs.png
+   :width: 120%
+   :align: center
+   :alt: dashboard-logs
+
+.. figure:: ../images/dashboard-jobs.png
+   :width: 120%
+   :align: center
+   :alt: dashboard-jobs

format.sh

@@ -1,8 +1,8 @@
 #!/usr/bin/env bash
 set -eo pipefail
 
-RUFF_VERSION=$(ruff --version | head -n 1 | awk '{print $2}')
-MYPY_VERSION=$(mypy --version | awk '{print $2}')
+RUFF_VERSION=$(poetry run ruff --version | head -n 1 | awk '{print $2}')
+MYPY_VERSION=$(poetry run mypy --version | awk '{print $2}')
 
 echo "ruff ver $RUFF_VERSION"
 echo "mypy ver $MYPY_VERSION"

konduktor/dashboard/README.md

@@ -0,0 +1,30 @@
+### Prereqs: kubectl is configured with remote machine/cluster
+
+# OPTION 1 (Automated Setup)
+
+To open the dashboard, run this inside the root konduktor directory:
+```
+./start_dashboard.sh
+```
+
+# OPTION 2 (Manual Setup)
+
+## 1. Apply kubernetes manifest
+Inside manifests directory (one with dashboard_deployment.yaml):
+```
+kubectl apply -f dashboard_deployment.yaml
+```
+Then, wait a minute or two for the pods to finish setup
+
+## 2. Port forward frontend in a terminal
+```
+kubectl port-forward svc/frontend 5173:5173 -n konduktor-dashboard
+```
+
+## 3. Port forward grafana in a terminal
+```
+kubectl port-forward svc/kube-prometheus-stack-grafana 3000:80 -n prometheus
+```
+
+## 4. Open dashboard at http://localhost:5173/
+

konduktor/dashboard/backend/main.py

@@ -0,0 +1,169 @@
+from typing import Any, Dict, List
+
+import socketio
+from fastapi import FastAPI, Request
+from fastapi.middleware.cors import CORSMiddleware
+from fastapi.responses import JSONResponse
+from kubernetes import client
+from kubernetes.client.exceptions import ApiException
+
+from konduktor import logging as konduktor_logging
+from konduktor.kube_client import batch_api, core_api, crd_api
+
+from .sockets import socketio as sio
+
+logger = konduktor_logging.get_logger2(__name__)
+
+# FastAPI app
+app = FastAPI()
+
+
+# CORS Configuration
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],  # Allow all origins
+    allow_credentials=True,
+    allow_methods=["*"],  # Allow all methods
+    allow_headers=["*"],  # Allow all headers
+)
+
+# Use Kubernetes API clients
+# Initialize BatchV1 and CoreV1 API (native kubernetes)
+batch_client = batch_api()
+core_client = core_api()
+# Initialize Kueue API
+crd_client = crd_api()
+
+
+@app.get("/")
+async def home():
+    return JSONResponse({"home": "/"})
+
+
+@app.delete("/deleteJob")
+async def delete_job(request: Request):
+    data = await request.json()
+    name = data.get("name", "")
+    namespace = data.get("namespace", "default")
+
+    try:
+        delete_options = client.V1DeleteOptions(propagation_policy="Background")
+
+        crd_client.delete_namespaced_custom_object(
+            group="kueue.x-k8s.io",
+            version="v1beta1",
+            namespace=namespace,
+            plural="workloads",
+            name=name,
+            body=delete_options,
+        )
+        logger.debug(f"Kueue Workload '{name}' deleted successfully.")
+
+        return JSONResponse({"success": True, "status": 200})
+
+    except ApiException as e:
+        logger.debug(f"Exception: {e}")
+        return JSONResponse({"error": str(e)}, status_code=e.status)
+
+
+@app.get("/getJobs")
+async def get_jobs():
+    rows = fetch_jobs()
+    return JSONResponse(rows)
+
+
+@app.get("/getNamespaces")
+async def get_namespaces():
+    try:
+        # Get the list of namespaces
+        namespaces = core_client.list_namespace()
+        # Extract the namespace names from the response
+        namespace_list = [ns.metadata.name for ns in namespaces.items]
+        return JSONResponse(namespace_list)
+    except ApiException as e:
+        logger.debug(f"Exception: {e}")
+        return JSONResponse({"error": str(e)}, status_code=e.status)
+
+
+@app.put("/updatePriority")
+async def update_priority(request: Request):
+    data = await request.json()
+    name = data.get("name", "")
+    namespace = data.get("namespace", "default")
+    priority = data.get("priority", 0)
+
+    try:
+        job = crd_client.get_namespaced_custom_object(
+            group="kueue.x-k8s.io",
+            version="v1beta1",
+            namespace=namespace,
+            plural="workloads",
+            name=name,
+        )
+
+        job["spec"]["priority"] = priority
+
+        crd_client.patch_namespaced_custom_object(
+            group="kueue.x-k8s.io",
+            version="v1beta1",
+            namespace=namespace,
+            plural="workloads",
+            name=name,
+            body=job,
+        )
+        return JSONResponse({"success": True, "status": 200})
+
+    except ApiException as e:
+        logger.debug(f"Exception: {e}")
+        return JSONResponse({"error": str(e)}, status_code=e.status)
+
+
+# Get a listing of workloads in kueue
+def fetch_jobs():
+    listing = crd_client.list_namespaced_custom_object(
+        group="kueue.x-k8s.io",
+        version="v1beta1",
+        namespace="default",
+        plural="workloads",
+    )
+
+    return format_workloads(listing)
+
+
+def format_workloads(listing: Dict[str, Any]) -> List[Dict[str, Any]]:
+    if not listing:
+        return []
+
+    res = []
+
+    for job in listing["items"]:
+        id = job["metadata"]["uid"]
+        name = job["metadata"]["name"]
+        created_at = job["metadata"]["creationTimestamp"]
+        namespace = job["metadata"]["namespace"]
+        localQueueName = job["spec"].get("queueName", "Unknown")
+        priority = job["spec"]["priority"]
+        active = job["spec"].get("active", 0)
+        status = "ADMITTED" if "admission" in job.get("status", {}) else "PENDING"
+
+        statusVal = 1 if "admission" in job.get("status", {}) else 0
+        order = (statusVal * 10) + priority
+
+        res.append(
+            {
+                "id": id,
+                "name": name,
+                "namespace": namespace,
+                "localQueueName": localQueueName,
+                "priority": priority,
+                "status": status,
+                "active": active,
+                "created_at": created_at,
+                "order": order,
+            }
+        )
+
+    return res
+
+
+app = socketio.ASGIApp(sio, app)