SQLDump Analyzer v3.0

A professional tool for analyzing SQL Server crash dump files with Azure AI integration.

Features

• Parse Dump Files: Read and analyze SQL Server minidump (.mdmp, .dmp) files
• Rich Knowledge Base: Preloaded explanations for:
  • Exception codes (ACCESS_VIOLATION, STACK_OVERFLOW, etc.)
  • SQL Server error codes (605, 823, 824, 17883, etc.)
  • Wait types (PAGEIOLATCH_, LCK_M_, CXPACKET, etc.)
  • Module/DLL meanings (sqlservr.exe, sqlmin.dll, etc.)
• WinDbg Integration: Automated analysis using cdb.exe (if installed)
• Azure AI Foundry Integration: Get intelligent, contextual explanations using AI

Quick Start

# Clone the repo
git clone https://github.com/chanman4821/SQLDump.git
cd SQLDump

# Install dependencies (optional - for AI features)
pip install -r requirements.txt

# Run the GUI
python run_gui.py

Screenshots

When you open a dump file, you'll see:

1. Overview Tab: Parsed dump info with explanations

   • File information
   • Exception details with "What this means" explanations
   • Common causes
   • Recommended next steps

2. WinDbg Analysis Tab: Detailed stack traces (requires Windows Debugging Tools)

3. AI Analysis Tab: Get intelligent analysis from Azure AI Foundry

   • Ask questions about your dump
   • Get root cause analysis
   • Receive actionable recommendations

4. Reference Panel:

   • Search SQL error codes by number
   • Look up wait types with explanations

Knowledge Base

The tool comes with a comprehensive preloaded knowledge base:

Exception Codes

Code	Name	What It Means
0xC0000005	ACCESS_VIOLATION	Memory access violation - SQL Server tried to access memory it shouldn't
0xC00000FD	STACK_OVERFLOW	Stack overflow from infinite recursion or deep call chains
0xE06D7363	CPP_EXCEPTION	C++ exception thrown and not caught
0x80000003	BREAKPOINT	Debugger breakpoint or internal assertion

SQL Server Errors

Error	Name	Category
605	Page Retrieval Failed	Database Corruption
823	OS I/O Error	I/O Error
824	Logical I/O Error	I/O Error
17883	Non-Yielding Scheduler	Scheduler

Wait Types

Wait Type	Category	Description
PAGEIOLATCH_*	I/O	Waiting to read pages from disk
WRITELOG	I/O	Transaction log writes
LCK_M_*	Locking	Lock acquisition waits
CXPACKET	Parallelism	Parallel query coordination

Azure AI Integration

To enable AI-powered analysis:

1. Go to Azure Portal → AI Foundry
2. Create a project and deploy a model (gpt-4o recommended)
3. In the app, go to Settings → Configure Azure AI
4. Enter your endpoint URL and API key

Or set environment variables:

$env:AZURE_AI_ENDPOINT = "https://your-endpoint.openai.azure.com/"
$env:AZURE_API_KEY = "your-api-key"
$env:AZURE_AI_MODEL = "gpt-4o"  # optional

Requirements

• Python 3.8+
• Windows (for WinDbg integration)
• Optional: Windows Debugging Tools (for detailed stack traces)
• Optional: Azure AI Foundry subscription (for AI analysis)

Installation

Windows Debugging Tools

For detailed stack trace analysis, install cdb.exe:

1. Download Windows SDK
2. During install, select "Debugging Tools for Windows"
3. Restart the app - it will auto-detect cdb.exe

Azure AI SDK

pip install azure-ai-inference azure-core

Usage

GUI Mode

python run_gui.py

CLI Mode

python run_cli.py path/to/dump.mdmp

Project Structure

SQLDump/
├── run_gui.py          # GUI entry point
├── run_cli.py          # CLI entry point
├── requirements.txt    # Python dependencies
├── sqldump/
│   ├── __init__.py     # Package init
│   ├── analyzer.py     # Core dump parsing
│   ├── windbg.py       # WinDbg/cdb.exe integration
│   ├── knowledge_base.py  # Preloaded reference data
│   ├── ai_analyzer.py  # Azure AI integration
│   ├── gui.py          # Main GUI application
│   └── errors.py       # Legacy error definitions

Contributing

Pull requests welcome! Please ensure:

• Code follows existing style
• New features include documentation
• Knowledge base additions include explanations

License

MIT License - Free to use and modify

Support

For SQL Server crash dump issues:

• Use this tool for initial analysis
• Check Microsoft documentation
• Contact Microsoft Support for critical issues

For tool issues:

• Open a GitHub issue
• Include Python version and error message