chore: vendor sglang v0.5.10 snapshot

This commit is contained in:
2026-04-24 12:29:36 +00:00
parent 78f0d15221
commit bded08301f
4308 changed files with 1200894 additions and 2 deletions

View File

@@ -0,0 +1,147 @@
# SGLang Performance Dashboard
A web-based dashboard for visualizing SGLang nightly test performance metrics.
## Features
- **Performance Trends**: View throughput, latency, and TTFT trends over time
- **Model Comparison**: Compare performance across different models and configurations
- **Filtering**: Filter by GPU configuration, model, variant, and batch size
- **Interactive Charts**: Zoom, pan, and hover for detailed metrics
- **Run History**: View recent benchmark runs with links to GitHub Actions
## Quick Start
### Option 1: Run with Local Server (Recommended)
For live data from GitHub Actions artifacts:
```bash
# Install requirements
pip install requests
# Run the server
python server.py --fetch-on-start
# Visit http://localhost:8000
```
The server provides:
- Automatic fetching of metrics from GitHub
- Caching to reduce API calls
- `/api/metrics` endpoint for the frontend
### Option 2: Fetch Data Manually
Use the fetch script to download metrics data:
```bash
# Fetch last 30 days of metrics
python fetch_metrics.py --output metrics_data.json
# Fetch a specific run
python fetch_metrics.py --run-id 21338741812 --output single_run.json
# Fetch only scheduled (nightly) runs
python fetch_metrics.py --scheduled-only --days 7
```
## GitHub Token
To download artifacts from GitHub, you need authentication:
1. **Using `gh` CLI** (recommended):
```bash
gh auth login
```
2. **Using environment variable**:
```bash
export GITHUB_TOKEN=your_token_here
```
Without a token, the dashboard will show run metadata but not detailed benchmark results.
## Data Structure
The metrics JSON has this structure:
```json
{
"run_id": "21338741812",
"run_date": "2026-01-25T22:24:02.090218+00:00",
"commit_sha": "5cdb391...",
"branch": "main",
"results": [
{
"gpu_config": "8-gpu-h200",
"partition": 0,
"model": "deepseek-ai/DeepSeek-V3.1",
"variant": "TP8+MTP",
"benchmarks": [
{
"batch_size": 1,
"input_len": 4096,
"output_len": 512,
"latency_ms": 2400.72,
"input_throughput": 21408.64,
"output_throughput": 231.74,
"overall_throughput": 1919.43,
"ttft_ms": 191.32,
"acc_length": 3.19
}
]
}
]
}
```
## Deployment
### GitHub Pages
The dashboard can be deployed to GitHub Pages for public access:
1. Copy the dashboard files to `docs/performance_dashboard/`
2. Enable GitHub Pages in repository settings
3. Set up a GitHub Action to periodically update metrics data
### Self-Hosted
For a self-hosted deployment with live data:
1. Set up a server running `server.py`
2. Configure a cron job or systemd timer to refresh data
3. Optionally put behind nginx/caddy for SSL
## Metrics Explained
- **Overall Throughput**: Total tokens (input + output) processed per second
- **Input Throughput**: Input tokens processed per second (prefill speed)
- **Output Throughput**: Output tokens generated per second (decode speed)
- **Latency**: End-to-end time to complete the request
- **TTFT**: Time to First Token - time until the first output token
- **Acc Length**: Acceptance length for speculative decoding (MTP variants)
## Contributing
To add support for new metrics or visualizations:
1. Update `fetch_metrics.py` if data collection needs changes
2. Modify `app.js` to add new chart types or filters
3. Update `index.html` for UI changes
## Troubleshooting
**No data displayed**
- Check browser console for errors
- Verify GitHub API is accessible
- Try running with `server.py --fetch-on-start`
**API rate limits**
- Use a GitHub token for higher limits
- The server caches data for 5 minutes
**Charts not rendering**
- Ensure Chart.js is loading from CDN
- Check for JavaScript errors in console

File diff suppressed because it is too large Load Diff

View File

@@ -0,0 +1,272 @@
#!/usr/bin/env python3
"""
Fetch and process SGLang nightly test metrics from GitHub Actions artifacts.
This script fetches consolidated metrics from GitHub Actions workflow runs
and outputs them as JSON for the performance dashboard.
Usage:
python fetch_metrics.py --output metrics_data.json
python fetch_metrics.py --output metrics_data.json --days 30
python fetch_metrics.py --output metrics_data.json --run-id 21338741812
"""
import argparse
import io
import json
import os
import sys
import zipfile
from datetime import datetime, timedelta, timezone
from pathlib import Path
from typing import Optional
import requests
GITHUB_REPO = "sgl-project/sglang"
WORKFLOW_NAME = "nightly-test-nvidia.yml"
ARTIFACT_PREFIX = "consolidated-metrics-"
def get_github_token() -> Optional[str]:
"""Get GitHub token from environment or gh CLI."""
# Check environment variable first
token = os.environ.get("GITHUB_TOKEN")
if token:
return token
# Try gh CLI
try:
import subprocess
result = subprocess.run(
["gh", "auth", "token"],
capture_output=True,
text=True,
check=True,
)
return result.stdout.strip()
except (subprocess.CalledProcessError, FileNotFoundError):
pass
return None
def get_headers(token: Optional[str]) -> dict:
"""Get request headers with optional authentication."""
headers = {
"Accept": "application/vnd.github.v3+json",
}
if token:
headers["Authorization"] = f"Bearer {token}"
return headers
def fetch_workflow_runs(
token: Optional[str],
days: int = 30,
event: Optional[str] = None,
) -> list:
"""Fetch completed workflow runs from GitHub Actions."""
url = f"https://api.github.com/repos/{GITHUB_REPO}/actions/workflows/{WORKFLOW_NAME}/runs"
params = {
"status": "completed",
"per_page": 100,
}
if event:
params["event"] = event
response = requests.get(url, headers=get_headers(token), params=params, timeout=30)
response.raise_for_status()
runs = response.json().get("workflow_runs", [])
# Filter by date
cutoff = datetime.now(timezone.utc) - timedelta(days=days)
runs = [
run
for run in runs
if datetime.fromisoformat(run["created_at"].replace("Z", "+00:00")) > cutoff
]
return runs
def fetch_run_artifacts(token: Optional[str], run_id: int) -> list:
"""Fetch artifacts for a specific workflow run."""
url = f"https://api.github.com/repos/{GITHUB_REPO}/actions/runs/{run_id}/artifacts"
response = requests.get(url, headers=get_headers(token), timeout=30)
response.raise_for_status()
return response.json().get("artifacts", [])
def download_artifact(token: Optional[str], artifact_id: int) -> Optional[bytes]:
"""Download an artifact by ID."""
if not token:
print(f"Warning: GitHub token required to download artifacts", file=sys.stderr)
return None
url = f"https://api.github.com/repos/{GITHUB_REPO}/actions/artifacts/{artifact_id}/zip"
headers = get_headers(token)
response = requests.get(url, headers=headers, allow_redirects=True, timeout=60)
if response.status_code == 200:
return response.content
print(
f"Failed to download artifact {artifact_id}: {response.status_code}",
file=sys.stderr,
)
return None
def extract_metrics_from_zip(zip_content: bytes) -> Optional[dict]:
"""Extract metrics JSON from a zip file."""
try:
with zipfile.ZipFile(io.BytesIO(zip_content)) as zf:
# Find the JSON file in the archive
json_files = [f for f in zf.namelist() if f.endswith(".json")]
if not json_files:
return None
with zf.open(json_files[0]) as f:
return json.load(f)
except (zipfile.BadZipFile, json.JSONDecodeError) as e:
print(f"Failed to extract metrics: {e}", file=sys.stderr)
return None
def fetch_metrics_for_run(token: Optional[str], run: dict) -> Optional[dict]:
"""Fetch metrics for a single workflow run."""
run_id = run["id"]
print(f"Fetching metrics for run {run_id}...", file=sys.stderr)
artifacts = fetch_run_artifacts(token, run_id)
# Find consolidated metrics artifact
metrics_artifact = None
for artifact in artifacts:
if artifact["name"].startswith(ARTIFACT_PREFIX):
metrics_artifact = artifact
break
if not metrics_artifact:
print(f"No consolidated metrics found for run {run_id}", file=sys.stderr)
return None
# Download and extract
zip_content = download_artifact(token, metrics_artifact["id"])
if not zip_content:
return None
metrics = extract_metrics_from_zip(zip_content)
if not metrics:
return None
# Ensure required fields are present
if "run_id" not in metrics:
metrics["run_id"] = str(run_id)
if "run_date" not in metrics:
metrics["run_date"] = run["created_at"]
if "commit_sha" not in metrics:
metrics["commit_sha"] = run["head_sha"]
if "branch" not in metrics:
metrics["branch"] = run["head_branch"]
return metrics
def fetch_single_run(token: Optional[str], run_id: int) -> Optional[dict]:
"""Fetch metrics for a single run by ID."""
url = f"https://api.github.com/repos/{GITHUB_REPO}/actions/runs/{run_id}"
response = requests.get(url, headers=get_headers(token), timeout=30)
response.raise_for_status()
run = response.json()
return fetch_metrics_for_run(token, run)
def main():
parser = argparse.ArgumentParser(
description="Fetch SGLang nightly test metrics from GitHub Actions"
)
parser.add_argument(
"--output",
"-o",
type=str,
default="metrics_data.json",
help="Output JSON file path",
)
parser.add_argument(
"--days",
type=int,
default=30,
help="Number of days to fetch (default: 30)",
)
parser.add_argument(
"--run-id",
type=int,
help="Fetch a specific run by ID",
)
parser.add_argument(
"--event",
type=str,
choices=["schedule", "workflow_dispatch", "push"],
help="Filter by trigger event type",
)
parser.add_argument(
"--scheduled-only",
action="store_true",
help="Only fetch scheduled (nightly) runs",
)
args = parser.parse_args()
token = get_github_token()
if not token:
print(
"Warning: No GitHub token found. Some features may be limited.",
file=sys.stderr,
)
print(
"Set GITHUB_TOKEN env var or login with 'gh auth login'",
file=sys.stderr,
)
all_metrics = []
if args.run_id:
# Fetch single run
metrics = fetch_single_run(token, args.run_id)
if metrics:
all_metrics.append(metrics)
else:
# Fetch multiple runs
event = "schedule" if args.scheduled_only else args.event
runs = fetch_workflow_runs(token, days=args.days, event=event)
print(f"Found {len(runs)} workflow runs", file=sys.stderr)
for run in runs:
metrics = fetch_metrics_for_run(token, run)
if metrics:
all_metrics.append(metrics)
# Sort by date descending
all_metrics.sort(key=lambda x: x.get("run_date", ""), reverse=True)
# Write output
output_path = Path(args.output)
with open(output_path, "w") as f:
json.dump(all_metrics, f, indent=2)
print(f"Wrote {len(all_metrics)} metrics records to {output_path}", file=sys.stderr)
if __name__ == "__main__":
main()

View File

@@ -0,0 +1,946 @@
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>SGLang Performance Dashboard</title>
<script src="https://cdn.jsdelivr.net/npm/chart.js"></script>
<script src="https://cdn.jsdelivr.net/npm/chartjs-adapter-date-fns"></script>
<link rel="preconnect" href="https://fonts.googleapis.com">
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
<link href="https://fonts.googleapis.com/css2?family=DM+Sans:ital,opsz,wght@0,9..40,300;0,9..40,400;0,9..40,500;0,9..40,600;0,9..40,700;1,9..40,400&family=JetBrains+Mono:wght@400;500;600;700&display=swap" rel="stylesheet">
<style>
:root {
--bg-primary: #0a0e17;
--bg-secondary: #111827;
--bg-tertiary: #1a2332;
--bg-elevated: #1e293b;
--text-primary: #e2e8f0;
--text-secondary: #94a3b8;
--text-muted: #64748b;
--border-color: #1e293b;
--border-subtle: rgba(148, 163, 184, 0.08);
--accent-cyan: #22d3ee;
--accent-cyan-dim: rgba(34, 211, 238, 0.15);
--accent-green: #34d399;
--accent-green-dim: rgba(52, 211, 153, 0.15);
--accent-amber: #fbbf24;
--accent-amber-dim: rgba(251, 191, 36, 0.15);
--accent-red: #f87171;
--accent-red-dim: rgba(248, 113, 113, 0.15);
--accent-violet: #a78bfa;
--accent-violet-dim: rgba(167, 139, 250, 0.15);
--glass-bg: rgba(17, 24, 39, 0.7);
--glass-border: rgba(148, 163, 184, 0.1);
--shadow-sm: 0 1px 2px rgba(0, 0, 0, 0.3);
--shadow-md: 0 4px 16px rgba(0, 0, 0, 0.3);
--shadow-lg: 0 12px 40px rgba(0, 0, 0, 0.4);
--radius-sm: 6px;
--radius-md: 10px;
--radius-lg: 14px;
--radius-xl: 20px;
}
* {
margin: 0;
padding: 0;
box-sizing: border-box;
}
body {
font-family: 'DM Sans', -apple-system, BlinkMacSystemFont, sans-serif;
background-color: var(--bg-primary);
color: var(--text-primary);
line-height: 1.6;
min-height: 100vh;
overflow-x: hidden;
}
/* Subtle grid background */
body::before {
content: '';
position: fixed;
inset: 0;
background-image:
linear-gradient(rgba(148, 163, 184, 0.03) 1px, transparent 1px),
linear-gradient(90deg, rgba(148, 163, 184, 0.03) 1px, transparent 1px);
background-size: 60px 60px;
pointer-events: none;
z-index: 0;
}
/* Ambient glow */
body::after {
content: '';
position: fixed;
top: -40%;
left: -20%;
width: 80%;
height: 80%;
background: radial-gradient(ellipse, rgba(34, 211, 238, 0.04) 0%, transparent 70%);
pointer-events: none;
z-index: 0;
}
.container {
max-width: 1480px;
margin: 0 auto;
padding: 24px 32px;
position: relative;
z-index: 1;
}
/* ---- Header ---- */
header {
display: flex;
justify-content: space-between;
align-items: center;
padding: 20px 0 24px;
margin-bottom: 28px;
border-bottom: 1px solid var(--border-subtle);
}
h1 {
font-size: 22px;
font-weight: 600;
display: flex;
align-items: center;
gap: 14px;
letter-spacing: -0.02em;
color: var(--text-primary);
}
.logo-mark {
width: 36px;
height: 36px;
border-radius: var(--radius-md);
background: linear-gradient(135deg, var(--accent-cyan-dim), rgba(167, 139, 250, 0.12));
border: 1px solid rgba(34, 211, 238, 0.2);
display: flex;
align-items: center;
justify-content: center;
flex-shrink: 0;
}
.logo-mark svg {
width: 20px;
height: 20px;
color: var(--accent-cyan);
}
h1 span.title-accent {
color: var(--accent-cyan);
}
.header-actions {
display: flex;
gap: 10px;
align-items: center;
}
.btn {
padding: 8px 18px;
border-radius: var(--radius-sm);
border: 1px solid var(--border-color);
background: var(--bg-secondary);
color: var(--text-secondary);
cursor: pointer;
font-size: 13px;
font-family: 'DM Sans', sans-serif;
font-weight: 500;
transition: all 0.2s ease;
text-decoration: none;
display: inline-flex;
align-items: center;
gap: 6px;
}
.btn:hover {
background: var(--bg-tertiary);
color: var(--text-primary);
border-color: var(--glass-border);
}
.btn svg {
width: 14px;
height: 14px;
}
.btn-primary {
background: linear-gradient(135deg, rgba(34, 211, 238, 0.15), rgba(34, 211, 238, 0.08));
border-color: rgba(34, 211, 238, 0.25);
color: var(--accent-cyan);
}
.btn-primary:hover {
background: linear-gradient(135deg, rgba(34, 211, 238, 0.25), rgba(34, 211, 238, 0.12));
border-color: rgba(34, 211, 238, 0.4);
}
/* ---- Stats Row ---- */
.stats-row {
display: grid;
grid-template-columns: repeat(auto-fit, minmax(200px, 1fr));
gap: 16px;
margin-bottom: 28px;
}
.stat-card {
background: var(--glass-bg);
backdrop-filter: blur(12px);
-webkit-backdrop-filter: blur(12px);
border-radius: var(--radius-lg);
border: 1px solid var(--glass-border);
padding: 20px 22px;
position: relative;
overflow: hidden;
transition: transform 0.2s ease, box-shadow 0.2s ease;
}
.stat-card:hover {
transform: translateY(-2px);
box-shadow: var(--shadow-md);
}
.stat-card::before {
content: '';
position: absolute;
top: 0;
left: 0;
right: 0;
height: 2px;
border-radius: var(--radius-lg) var(--radius-lg) 0 0;
}
.stat-card:nth-child(1)::before { background: linear-gradient(90deg, var(--accent-cyan), transparent); }
.stat-card:nth-child(2)::before { background: linear-gradient(90deg, var(--accent-violet), transparent); }
.stat-card:nth-child(3)::before { background: linear-gradient(90deg, var(--accent-green), transparent); }
.stat-card .label {
font-size: 11px;
color: var(--text-muted);
text-transform: uppercase;
letter-spacing: 0.08em;
font-weight: 500;
margin-bottom: 8px;
}
.stat-card .value {
font-family: 'JetBrains Mono', monospace;
font-size: 28px;
font-weight: 700;
color: var(--text-primary);
letter-spacing: -0.02em;
}
.stat-card .change {
font-size: 12px;
margin-top: 6px;
font-weight: 500;
}
.stat-card .change.positive { color: var(--accent-green); }
.stat-card .change.negative { color: var(--accent-red); }
/* ---- Filters ---- */
.filters {
display: flex;
gap: 14px;
flex-wrap: wrap;
margin-bottom: 28px;
padding: 18px 22px;
background: var(--glass-bg);
backdrop-filter: blur(12px);
-webkit-backdrop-filter: blur(12px);
border-radius: var(--radius-lg);
border: 1px solid var(--glass-border);
align-items: flex-end;
}
.filter-group {
display: flex;
flex-direction: column;
gap: 6px;
flex: 1;
min-width: 160px;
}
.filter-group label {
font-size: 10px;
color: var(--text-muted);
font-weight: 600;
text-transform: uppercase;
letter-spacing: 0.1em;
}
select {
padding: 9px 32px 9px 14px;
border-radius: var(--radius-sm);
border: 1px solid var(--border-color);
background: var(--bg-tertiary);
color: var(--text-primary);
font-size: 13px;
font-family: 'DM Sans', sans-serif;
font-weight: 500;
cursor: pointer;
transition: all 0.15s ease;
appearance: none;
-webkit-appearance: none;
background-image: url("data:image/svg+xml,%3Csvg xmlns='http://www.w3.org/2000/svg' width='12' height='12' viewBox='0 0 24 24' fill='none' stroke='%2394a3b8' stroke-width='2' stroke-linecap='round' stroke-linejoin='round'%3E%3Cpath d='M6 9l6 6 6-6'/%3E%3C/svg%3E");
background-repeat: no-repeat;
background-position: right 10px center;
width: 100%;
}
select:hover {
border-color: rgba(148, 163, 184, 0.2);
}
select:focus {
outline: none;
border-color: rgba(34, 211, 238, 0.4);
box-shadow: 0 0 0 3px rgba(34, 211, 238, 0.08);
}
/* ---- Metric Tabs ---- */
.tabs {
display: flex;
gap: 2px;
margin-bottom: 24px;
padding: 4px;
background: var(--bg-secondary);
border-radius: var(--radius-md);
border: 1px solid var(--border-subtle);
width: fit-content;
}
.tab {
padding: 9px 18px;
cursor: pointer;
border-radius: var(--radius-sm);
background: transparent;
color: var(--text-muted);
border: none;
transition: all 0.2s ease;
font-weight: 500;
font-size: 13px;
font-family: 'DM Sans', sans-serif;
white-space: nowrap;
}
.tab:hover {
color: var(--text-secondary);
background: rgba(148, 163, 184, 0.05);
}
.tab.active {
background: var(--bg-tertiary);
color: var(--accent-cyan);
box-shadow: var(--shadow-sm);
}
/* ---- Chart Cards ---- */
.chart-card {
background: var(--glass-bg);
backdrop-filter: blur(12px);
-webkit-backdrop-filter: blur(12px);
border-radius: var(--radius-lg);
border: 1px solid var(--glass-border);
padding: 24px;
}
.chart-card h3 {
font-size: 15px;
font-weight: 600;
margin-bottom: 20px;
color: var(--text-primary);
display: flex;
align-items: center;
gap: 10px;
letter-spacing: -0.01em;
}
.chart-card h3::before {
content: '';
width: 3px;
height: 18px;
background: var(--accent-cyan);
border-radius: 2px;
flex-shrink: 0;
}
.chart-container {
position: relative;
height: 320px;
}
.metric-section {
margin-bottom: 24px;
}
.batch-charts-container {
display: grid;
grid-template-columns: repeat(auto-fit, minmax(420px, 1fr));
gap: 18px;
}
.batch-chart-wrapper {
background: var(--bg-tertiary);
border-radius: var(--radius-md);
padding: 16px;
border: 1px solid var(--border-subtle);
}
.batch-chart-title {
font-family: 'JetBrains Mono', monospace;
font-size: 12px;
font-weight: 500;
color: var(--text-muted);
margin-bottom: 10px;
text-align: center;
text-transform: uppercase;
letter-spacing: 0.06em;
}
.charts-grid {
display: grid;
grid-template-columns: repeat(auto-fit, minmax(600px, 1fr));
gap: 24px;
}
/* ---- Data Table ---- */
.data-table {
width: 100%;
border-collapse: separate;
border-spacing: 0;
margin-top: 20px;
}
.data-table th {
padding: 10px 16px;
text-align: left;
font-size: 10px;
font-weight: 600;
color: var(--text-muted);
text-transform: uppercase;
letter-spacing: 0.08em;
border-bottom: 1px solid var(--border-color);
background: transparent;
}
.data-table td {
padding: 12px 16px;
text-align: left;
border-bottom: 1px solid var(--border-subtle);
font-size: 13px;
color: var(--text-secondary);
}
.data-table tbody tr {
transition: background 0.15s ease;
}
.data-table tbody tr:hover {
background: rgba(148, 163, 184, 0.04);
}
.data-table td code {
font-family: 'JetBrains Mono', monospace;
font-size: 12px;
color: var(--accent-cyan);
background: var(--accent-cyan-dim);
padding: 2px 8px;
border-radius: 4px;
}
.run-link {
font-family: 'JetBrains Mono', monospace;
font-size: 12px;
color: var(--accent-cyan);
text-decoration: none;
transition: color 0.15s;
}
.run-link:hover {
color: #67e8f9;
text-decoration: underline;
}
.model-badge {
display: inline-block;
padding: 3px 10px;
border-radius: 20px;
font-size: 11px;
font-weight: 500;
background: var(--accent-violet-dim);
color: var(--accent-violet);
border: 1px solid rgba(167, 139, 250, 0.15);
}
/* ---- Loading ---- */
.loading {
display: flex;
flex-direction: column;
justify-content: center;
align-items: center;
min-height: 400px;
gap: 20px;
color: var(--text-muted);
}
.spinner {
width: 36px;
height: 36px;
border: 2px solid var(--border-color);
border-top-color: var(--accent-cyan);
border-radius: 50%;
animation: spin 0.8s linear infinite;
}
@keyframes spin {
to { transform: rotate(360deg); }
}
.loading-text {
font-size: 13px;
font-weight: 500;
color: var(--text-muted);
}
/* ---- Error ---- */
.error {
background: var(--accent-red-dim);
border: 1px solid rgba(248, 113, 113, 0.2);
border-radius: var(--radius-lg);
padding: 28px;
text-align: center;
color: var(--accent-red);
}
.error h3 {
margin-bottom: 8px;
font-size: 16px;
}
.error p {
font-size: 13px;
color: rgba(248, 113, 113, 0.8);
}
.no-data {
text-align: center;
padding: 60px 20px;
color: var(--text-muted);
font-size: 14px;
}
.no-data h3 {
margin-bottom: 8px;
}
/* ---- Footer ---- */
footer {
margin-top: 48px;
padding: 28px 0;
border-top: 1px solid var(--border-subtle);
text-align: center;
color: var(--text-muted);
font-size: 13px;
}
footer a {
color: var(--text-secondary);
text-decoration: none;
transition: color 0.15s;
}
footer a:hover {
color: var(--accent-cyan);
}
/* ---- Login Overlay ---- */
.login-overlay {
position: fixed;
inset: 0;
background-color: var(--bg-primary);
display: flex;
justify-content: center;
align-items: center;
z-index: 1000;
overflow: hidden;
}
.login-overlay::before {
content: '';
position: absolute;
inset: 0;
background-image:
linear-gradient(rgba(148, 163, 184, 0.03) 1px, transparent 1px),
linear-gradient(90deg, rgba(148, 163, 184, 0.03) 1px, transparent 1px);
background-size: 60px 60px;
pointer-events: none;
}
.login-overlay::after {
content: '';
position: absolute;
top: 50%;
left: 50%;
transform: translate(-50%, -50%);
width: 600px;
height: 600px;
background: radial-gradient(ellipse, rgba(34, 211, 238, 0.06) 0%, transparent 70%);
pointer-events: none;
}
.login-card {
background: var(--glass-bg);
backdrop-filter: blur(20px);
-webkit-backdrop-filter: blur(20px);
border: 1px solid var(--glass-border);
border-radius: var(--radius-xl);
padding: 44px 40px;
width: 100%;
max-width: 400px;
box-shadow: var(--shadow-lg);
position: relative;
z-index: 1;
animation: loginSlideUp 0.5s ease-out;
}
@keyframes loginSlideUp {
from {
opacity: 0;
transform: translateY(20px);
}
to {
opacity: 1;
transform: translateY(0);
}
}
.login-icon {
text-align: center;
margin-bottom: 20px;
}
.login-icon-wrapper {
width: 56px;
height: 56px;
margin: 0 auto;
border-radius: var(--radius-lg);
background: linear-gradient(135deg, var(--accent-cyan-dim), rgba(167, 139, 250, 0.12));
border: 1px solid rgba(34, 211, 238, 0.2);
display: flex;
align-items: center;
justify-content: center;
}
.login-icon-wrapper svg {
width: 24px;
height: 24px;
color: var(--accent-cyan);
}
.login-card h2 {
font-size: 20px;
font-weight: 600;
margin-bottom: 6px;
text-align: center;
letter-spacing: -0.02em;
}
.login-card .login-subtitle {
font-size: 13px;
color: var(--text-muted);
text-align: center;
margin-bottom: 28px;
}
.login-card .form-group {
margin-bottom: 18px;
}
.login-card .form-group label {
display: block;
font-size: 12px;
color: var(--text-muted);
margin-bottom: 7px;
font-weight: 500;
}
.login-card .form-group input {
width: 100%;
padding: 11px 14px;
border-radius: var(--radius-sm);
border: 1px solid var(--border-color);
background: var(--bg-tertiary);
color: var(--text-primary);
font-size: 14px;
font-family: 'DM Sans', sans-serif;
outline: none;
transition: all 0.2s ease;
}
.login-card .form-group input:focus {
border-color: rgba(34, 211, 238, 0.4);
box-shadow: 0 0 0 3px rgba(34, 211, 238, 0.08);
}
.login-card .form-group input::placeholder {
color: var(--text-muted);
}
.login-card .login-btn {
width: 100%;
padding: 11px 16px;
border-radius: var(--radius-sm);
border: 1px solid rgba(34, 211, 238, 0.3);
background: linear-gradient(135deg, rgba(34, 211, 238, 0.15), rgba(34, 211, 238, 0.08));
color: var(--accent-cyan);
font-size: 14px;
font-family: 'DM Sans', sans-serif;
font-weight: 600;
cursor: pointer;
transition: all 0.2s ease;
margin-top: 6px;
}
.login-card .login-btn:hover {
background: linear-gradient(135deg, rgba(34, 211, 238, 0.25), rgba(34, 211, 238, 0.12));
border-color: rgba(34, 211, 238, 0.5);
}
.login-card .login-btn:disabled {
opacity: 0.5;
cursor: not-allowed;
}
.login-error {
color: var(--accent-red);
font-size: 13px;
text-align: center;
margin-top: 14px;
min-height: 20px;
}
/* ---- Entrance Animations ---- */
@keyframes fadeInUp {
from {
opacity: 0;
transform: translateY(12px);
}
to {
opacity: 1;
transform: translateY(0);
}
}
.animate-in {
animation: fadeInUp 0.4s ease-out both;
}
.animate-delay-1 { animation-delay: 0.05s; }
.animate-delay-2 { animation-delay: 0.1s; }
.animate-delay-3 { animation-delay: 0.15s; }
.animate-delay-4 { animation-delay: 0.2s; }
.animate-delay-5 { animation-delay: 0.25s; }
.animate-delay-6 { animation-delay: 0.3s; }
/* ---- Responsive ---- */
@media (max-width: 768px) {
.container {
padding: 16px;
}
header {
flex-direction: column;
gap: 16px;
align-items: flex-start;
}
.filters {
padding: 14px;
}
.filter-group {
min-width: 140px;
}
.tabs {
overflow-x: auto;
-webkit-overflow-scrolling: touch;
}
.batch-charts-container {
grid-template-columns: 1fr;
}
.login-card {
margin: 16px;
padding: 32px 24px;
}
.stat-card .value {
font-size: 22px;
}
}
/* ---- Scrollbar ---- */
::-webkit-scrollbar {
width: 6px;
height: 6px;
}
::-webkit-scrollbar-track {
background: transparent;
}
::-webkit-scrollbar-thumb {
background: var(--border-color);
border-radius: 3px;
}
::-webkit-scrollbar-thumb:hover {
background: var(--text-muted);
}
</style>
</head>
<body>
<!-- Login overlay -->
<div id="login-overlay" class="login-overlay">
<div class="login-card">
<div class="login-icon">
<div class="login-icon-wrapper">
<svg viewBox="0 0 24 24" fill="none" xmlns="http://www.w3.org/2000/svg">
<rect x="3" y="11" width="18" height="11" rx="2" ry="2" stroke="currentColor" stroke-width="2"/>
<path d="M7 11V7a5 5 0 0 1 10 0v4" stroke="currentColor" stroke-width="2" stroke-linecap="round"/>
<circle cx="12" cy="16" r="1.5" fill="currentColor"/>
</svg>
</div>
</div>
<h2>SGLang Performance Dashboard</h2>
<p class="login-subtitle">Enter your credentials to access the dashboard</p>
<form id="login-form" onsubmit="return handleLogin(event)">
<div class="form-group">
<label for="login-username">Username</label>
<input type="text" id="login-username" name="username" autocomplete="username" placeholder="Enter username" required autofocus>
</div>
<div class="form-group">
<label for="login-password">Password</label>
<input type="password" id="login-password" name="password" autocomplete="current-password" placeholder="Enter password" required>
</div>
<button type="submit" class="login-btn" id="login-btn">Sign In</button>
</form>
<div id="login-error" class="login-error"></div>
</div>
</div>
<div class="container" id="dashboard-container" style="display: none;">
<header class="animate-in">
<h1>
<div class="logo-mark">
<svg viewBox="0 0 24 24" fill="none" xmlns="http://www.w3.org/2000/svg">
<path d="M12 2L2 7L12 12L22 7L12 2Z" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"/>
<path d="M2 17L12 22L22 17" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"/>
<path d="M2 12L12 17L22 12" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"/>
</svg>
</div>
<span><span class="title-accent">SGLang</span> Performance Dashboard</span>
</h1>
<div class="header-actions">
<button class="btn" onclick="refreshData()">
<svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><polyline points="23 4 23 10 17 10"/><path d="M20.49 15a9 9 0 1 1-2.12-9.36L23 10"/></svg>
Refresh
</button>
<a href="https://github.com/sgl-project/sglang/actions/workflows/nightly-test-nvidia.yml?query=event%3Aschedule" target="_blank" class="btn">
<svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><path d="M18 13v6a2 2 0 0 1-2 2H5a2 2 0 0 1-2-2V8a2 2 0 0 1 2-2h6"/><polyline points="15 3 21 3 21 9"/><line x1="10" y1="14" x2="21" y2="3"/></svg>
Workflow
</a>
</div>
</header>
<div id="loading" class="loading">
<div class="spinner"></div>
<div class="loading-text">Loading performance data...</div>
</div>
<div id="content" style="display: none;">
<div class="stats-row animate-in animate-delay-1" id="stats-row"></div>
<div class="filters animate-in animate-delay-2">
<div class="filter-group">
<label>GPU Configuration</label>
<select id="gpu-filter" onchange="handleGpuFilterChange()">
</select>
</div>
<div class="filter-group">
<label>Model</label>
<select id="model-filter" onchange="handleModelFilterChange(this.value)">
</select>
</div>
<div class="filter-group">
<label>Variant</label>
<select id="variant-filter" onchange="updateCharts()">
<option value="all">All Variants</option>
</select>
</div>
<div class="filter-group">
<label>Input / Output Length</label>
<select id="io-len-filter" onchange="updateCharts()">
<option value="all">All Lengths</option>
</select>
</div>
<div class="filter-group">
<label>Batch Size</label>
<select id="batch-filter" onchange="updateCharts()">
<option value="all">All Batch Sizes</option>
</select>
</div>
</div>
<div class="tabs animate-in animate-delay-3" id="metric-tabs"></div>
<div class="metric-section animate-in animate-delay-4">
<div class="chart-card">
<h3 id="metric-title">Overall Throughput (tokens/sec)</h3>
<div class="batch-charts-container" id="charts-container">
</div>
</div>
</div>
<div class="chart-card animate-in animate-delay-5" style="margin-top: 24px;">
<h3>Recent Benchmark Runs</h3>
<table class="data-table" id="runs-table">
<thead>
<tr>
<th>Date</th>
<th>Run ID</th>
<th>Commit</th>
<th>Branch</th>
<th>Models Tested</th>
</tr>
</thead>
<tbody id="runs-table-body">
</tbody>
</table>
</div>
</div>
<div id="error" class="error" style="display: none;">
<h3>Failed to load performance data</h3>
<p id="error-message"></p>
</div>
<footer class="animate-in animate-delay-6">
<p>
SGLang Performance Dashboard &mdash;
<a href="https://github.com/sgl-project/sglang" target="_blank">GitHub</a> &middot;
<a href="https://docs.sglang.io/" target="_blank">Documentation</a>
</p>
</footer>
</div>
<script src="app.js"></script>
</body>
</html>

View File

@@ -0,0 +1,422 @@
#!/usr/bin/env python3
"""
Simple development server for the SGLang Performance Dashboard.
This server:
1. Serves the static HTML/JS files
2. Provides an API endpoint to fetch metrics from GitHub
3. Caches metrics data to reduce API calls
Usage:
python server.py
python server.py --port 8080
python server.py --host 0.0.0.0 # Allow external access
python server.py --fetch-on-start
python server.py --username admin --password secret # Enable authentication
DASHBOARD_USERNAME=admin DASHBOARD_PASSWORD=secret python server.py # Via env vars
python server.py --refresh-interval 12 # Auto-refresh data every 12 hours
"""
import argparse
import hashlib
import hmac
import http.server
import io
import json
import os
import secrets
import socketserver
import threading
import time
import zipfile
from datetime import datetime, timedelta, timezone
from pathlib import Path
from urllib.parse import urlparse
import requests
GITHUB_REPO = "sgl-project/sglang"
WORKFLOW_NAME = "nightly-test-nvidia.yml"
ARTIFACT_PREFIX = "consolidated-metrics-"
# Cache for metrics data with thread-safe lock
cache_lock = threading.Lock()
metrics_cache = {
"data": [],
"last_updated": None,
"updating": False,
}
CACHE_TTL = 300 # 5 minutes
REQUEST_TIMEOUT = 30 # seconds
# Authentication configuration (set via CLI flags)
auth_config = {
"enabled": False,
"username": None,
"password_hash": None, # SHA-256 hash of the password
"active_tokens": {}, # token -> expiry timestamp
}
auth_lock = threading.Lock()
AUTH_TOKEN_TTL = 3600 # 1 hour
def hash_password(password):
"""Hash a password using SHA-256 for constant-time comparison."""
return hashlib.sha256(password.encode("utf-8")).hexdigest()
def create_auth_token():
"""Create a new session token."""
token = secrets.token_hex(32)
with auth_lock:
# Clean up expired tokens
now = time.time()
auth_config["active_tokens"] = {
t: exp for t, exp in auth_config["active_tokens"].items() if exp > now
}
auth_config["active_tokens"][token] = now + AUTH_TOKEN_TTL
return token
def verify_auth_token(token):
"""Verify a session token is valid and not expired."""
if not token:
return False
with auth_lock:
expiry = auth_config["active_tokens"].get(token)
if expiry and expiry > time.time():
return True
# Remove expired token
auth_config["active_tokens"].pop(token, None)
return False
def get_github_token():
"""Get GitHub token from environment or gh CLI."""
token = os.environ.get("GITHUB_TOKEN")
if token:
return token
try:
import subprocess
result = subprocess.run(
["gh", "auth", "token"],
capture_output=True,
text=True,
check=True,
)
return result.stdout.strip()
except (subprocess.CalledProcessError, FileNotFoundError):
pass
return None
def fetch_metrics_from_github(days=30):
"""Fetch metrics from GitHub Actions artifacts."""
token = get_github_token()
headers = {"Accept": "application/vnd.github.v3+json"}
if token:
headers["Authorization"] = f"Bearer {token}"
# Get workflow runs - only scheduled (nightly) runs, not workflow_dispatch
url = f"https://api.github.com/repos/{GITHUB_REPO}/actions/workflows/{WORKFLOW_NAME}/runs"
params = {"status": "completed", "per_page": 50, "event": "schedule"}
try:
response = requests.get(
url, headers=headers, params=params, timeout=REQUEST_TIMEOUT
)
if not response.ok:
print(f"Failed to fetch workflow runs: {response.status_code}")
return []
except requests.exceptions.RequestException as e:
print(f"Network error fetching workflow runs: {e}")
return []
runs = response.json().get("workflow_runs", [])
# Filter by date
cutoff = datetime.now(timezone.utc) - timedelta(days=days)
runs = [
run
for run in runs
if datetime.fromisoformat(run["created_at"].replace("Z", "+00:00")) > cutoff
]
all_metrics = []
for run in runs[:20]: # Limit to 20 most recent
run_id = run["id"]
# Get artifacts
artifacts_url = f"https://api.github.com/repos/{GITHUB_REPO}/actions/runs/{run_id}/artifacts"
try:
artifacts_resp = requests.get(
artifacts_url, headers=headers, timeout=REQUEST_TIMEOUT
)
if not artifacts_resp.ok:
continue
except requests.exceptions.RequestException as e:
print(f"Network error fetching artifacts for run {run_id}: {e}")
continue
artifacts = artifacts_resp.json().get("artifacts", [])
# Find consolidated metrics
for artifact in artifacts:
if artifact["name"].startswith(ARTIFACT_PREFIX):
if not token:
# Without token, we can't download - return metadata only
all_metrics.append(
{
"run_id": str(run_id),
"run_date": run["created_at"],
"commit_sha": run["head_sha"],
"branch": run["head_branch"],
"results": [],
}
)
break
# Download artifact
download_url = f"https://api.github.com/repos/{GITHUB_REPO}/actions/artifacts/{artifact['id']}/zip"
try:
download_resp = requests.get(
download_url,
headers=headers,
allow_redirects=True,
timeout=REQUEST_TIMEOUT,
)
except requests.exceptions.RequestException as e:
print(f"Network error downloading artifact: {e}")
break
if download_resp.ok:
try:
with zipfile.ZipFile(io.BytesIO(download_resp.content)) as zf:
json_files = [
f for f in zf.namelist() if f.endswith(".json")
]
if json_files:
with zf.open(json_files[0]) as f:
metrics = json.load(f)
# Ensure required fields
metrics.setdefault("run_id", str(run_id))
metrics.setdefault("run_date", run["created_at"])
metrics.setdefault("commit_sha", run["head_sha"])
metrics.setdefault("branch", run["head_branch"])
all_metrics.append(metrics)
except (zipfile.BadZipFile, json.JSONDecodeError) as e:
print(f"Failed to process artifact: {e}")
break
return all_metrics
def update_cache_async():
"""Update the metrics cache in background with thread safety."""
with cache_lock:
if metrics_cache["updating"]:
return
metrics_cache["updating"] = True
try:
data = fetch_metrics_from_github()
with cache_lock:
metrics_cache["data"] = data
metrics_cache["last_updated"] = time.time()
print(f"Cache updated with {len(data)} metrics records")
finally:
with cache_lock:
metrics_cache["updating"] = False
def start_periodic_refresh(interval_hours):
"""Start a background thread that refreshes the cache periodically."""
interval_seconds = interval_hours * 3600
def refresh_loop():
while True:
time.sleep(interval_seconds)
print(f"Periodic refresh triggered (every {interval_hours}h)")
update_cache_async()
thread = threading.Thread(target=refresh_loop, daemon=True)
thread.start()
print(f"Periodic refresh enabled: every {interval_hours} hours")
class DashboardHandler(http.server.SimpleHTTPRequestHandler):
"""HTTP request handler for the dashboard."""
def __init__(self, *args, directory=None, **kwargs):
super().__init__(*args, directory=directory, **kwargs)
def _send_json(self, data, status=200):
"""Send a JSON response."""
self.send_response(status)
self.send_header("Content-Type", "application/json")
self.send_header("Access-Control-Allow-Origin", "*")
self.end_headers()
self.wfile.write(json.dumps(data).encode())
def _check_auth(self):
"""Check if request is authenticated. Returns True if OK, sends 401 and returns False otherwise."""
if not auth_config["enabled"]:
return True
auth_header = self.headers.get("Authorization", "")
if auth_header.startswith("Bearer "):
token = auth_header[7:]
if verify_auth_token(token):
return True
self._send_json({"error": "Unauthorized"}, status=401)
return False
def do_GET(self):
parsed = urlparse(self.path)
# Prevent directory traversal attacks
if ".." in parsed.path or parsed.path.startswith("//"):
self.send_error(400, "Invalid path")
return
if parsed.path == "/api/auth-check":
self.handle_auth_check()
elif parsed.path == "/api/metrics":
if self._check_auth():
self.handle_metrics_api(parsed)
elif parsed.path == "/api/refresh":
if self._check_auth():
self.handle_refresh_api()
else:
super().do_GET()
def do_POST(self):
parsed = urlparse(self.path)
if parsed.path == "/api/login":
self.handle_login()
else:
self.send_error(404, "Not Found")
def handle_auth_check(self):
"""Tell the frontend whether authentication is required."""
self._send_json({"auth_required": auth_config["enabled"]})
def handle_login(self):
"""Validate username/password and return a session token."""
content_length = int(self.headers.get("Content-Length", 0))
if content_length == 0 or content_length > 4096:
self._send_json({"error": "Invalid request"}, status=400)
return
try:
body = json.loads(self.rfile.read(content_length))
except (json.JSONDecodeError, ValueError):
self._send_json({"error": "Invalid JSON"}, status=400)
return
username = body.get("username", "")
password = body.get("password", "")
if hmac.compare_digest(
username, auth_config["username"]
) and hmac.compare_digest(
hash_password(password), auth_config["password_hash"]
):
token = create_auth_token()
self._send_json({"token": token})
else:
self._send_json({"error": "Invalid username or password"}, status=401)
def handle_metrics_api(self, parsed):
"""Handle /api/metrics endpoint."""
# Check cache with thread safety
with cache_lock:
cache_valid = (
metrics_cache["last_updated"]
and time.time() - metrics_cache["last_updated"] < CACHE_TTL
)
data = metrics_cache["data"].copy()
if not cache_valid:
# Trigger background update
threading.Thread(target=update_cache_async, daemon=True).start()
self._send_json(data)
def handle_refresh_api(self):
"""Handle /api/refresh endpoint."""
threading.Thread(target=update_cache_async, daemon=True).start()
self._send_json({"status": "refreshing"})
def log_message(self, format, *args):
"""Custom log format."""
print(f"[{self.log_date_time_string()}] {args[0]}")
def main():
parser = argparse.ArgumentParser(description="SGLang Performance Dashboard Server")
parser.add_argument("--port", type=int, default=8000, help="Port to serve on")
parser.add_argument(
"--host",
default="127.0.0.1",
help="Host to bind to (use 0.0.0.0 for external access)",
)
parser.add_argument(
"--fetch-on-start", action="store_true", help="Fetch metrics on startup"
)
parser.add_argument(
"--refresh-interval",
type=float,
default=12,
help="Auto-refresh interval in hours (default: 12, set to 0 to disable)",
)
parser.add_argument(
"--username",
default=os.environ.get("DASHBOARD_USERNAME"),
help="Username for dashboard authentication (or set DASHBOARD_USERNAME env var)",
)
parser.add_argument(
"--password",
default=os.environ.get("DASHBOARD_PASSWORD"),
help="Password for dashboard authentication (or set DASHBOARD_PASSWORD env var)",
)
args = parser.parse_args()
# Configure authentication if both username and password are provided
if args.username and args.password:
auth_config["enabled"] = True
auth_config["username"] = args.username
auth_config["password_hash"] = hash_password(args.password)
print(f"Authentication enabled for user: {args.username}")
elif args.username or args.password:
parser.error("Both --username and --password must be provided together")
# Change to dashboard directory
dashboard_dir = Path(__file__).parent
os.chdir(dashboard_dir)
if args.fetch_on_start:
print("Fetching initial metrics data...")
update_cache_async()
if args.refresh_interval > 0:
start_periodic_refresh(args.refresh_interval)
handler = lambda *a, **kw: DashboardHandler(*a, directory=str(dashboard_dir), **kw)
with socketserver.TCPServer((args.host, args.port), handler) as httpd:
print(f"Serving dashboard at http://{args.host}:{args.port}")
print("Press Ctrl+C to stop")
try:
httpd.serve_forever()
except KeyboardInterrupt:
print("\nShutting down...")
if __name__ == "__main__":
main()