LineageKnot: Self-Healing Dependency Graph Sentinel

LineageKnot is an active infrastructure dependency sentinel that translates relational table schemas into a self-healing graph engine. It traces microservice lineages, detects circular cyclical dependencies, and dynamically repairs loop anomalies under strict transactional isolation.

---

Repository Structure

The project has been organized according to professional data-engineering repository standards:

lineage-knot/
├── .env.example                # Template for database configuration parameters
├── .gitignore                  # Git tracking exclusion list
├── LICENSE                     # MIT License
├── README.md                   # Project documentation
├── run_sentinel.ps1            # Master automation runner (PowerShell Core)
└── sql/
    ├── schema/
    │   └── 01_system_dependencies.sql     # Database structures and audit logging tables
    ├── seeds/
    │   └── 01_seed_dependencies.sql      # Mock dependency topology seeds (8 microservices)
    └── procedures/
        └── 01_sentinel_views_procs.sql    # Recursive CTE Graph View and Self-Healing Stored Proc

---

Architectural Features

1. Cycle Detection (Recursive CTE): Discovers downstream systems starting from terminal root endpoints (target_node IS NULL) utilizing a recursive Common Table Expression (CTE). It detects cycles in real-time, measuring path depth, and flags loops instantly if a node is visited twice in a lineage path.
2. Transaction-Safe (TOCTOU Proof): The scanning, anomaly isolation, and updating occur inside a single stored procedure (silver.sp_self_heal_dependencies) under a SERIALIZABLE transaction isolation level. This locks the topology table, preventing race conditions or concurrency conflicts.
3. Structure Preservation & Auditing: Instead of deleting connections or setting them to NULL, loops are broken by setting the link status to DEGRADED_HEALED. The view automatically filters out degraded paths, resolving the cycle while preserving structural history in the database.
4. Automated Audit Logs: Every healing event logs the corrupted node, broken upstream target, timestamp, and a SHA-256 confirmation code in silver.dependency_healing_log.

---

Setup & Prerequisites

Prerequisites

• Microsoft SQL Server 2022 (Running locally or in a Docker container)
• PowerShell Core (7.x recommended or Windows PowerShell 5.1+)
• Docker Desktop (if deploying SQL Server via containerization)

Environment Configuration

1. Copy the template configuration file:

   cp .env.example .env

2. Open .env and set your SQL Server system administrator (sa) password:

   MSSQL_SA_PASSWORD=YourSecure_Password123!

---

Getting Started

To execute the entire end-to-end pipeline, run the master runner script:

powershell -File run_sentinel.ps1

Execution Telemetry Output Example

==========================================================================
             LINEAGEKNOT: AUTONOMOUS SELF-HEALING ENGINE                  
==========================================================================

[1/4] Generating Database Structures (DDL)...
STATUS: Schema and log tables created.

[2/4] Ingesting Seed Data Package (DML)...
STATUS: Ingested 8 dependency records.

[3/4] Compiling Sentinel View & Stored Procedure...
STATUS: Views and stored procedures compiled.

[4/4] Starting Graph Scan & Autonomous Stored Proc Execution...
ANOMALY DETECTED!
Loop Lineage Trace : PortalService -> ServiceA -> ServiceC -> ServiceB -> ServiceA
Isolated Malicious Node: ServiceA (closes loop back to ServiceB)
STATUS: Autonomous self-healing transaction completed successfully.
Confirmation Code: CONFIRMATION_CODE_SHA256_BE1C522E0C4EBCD016322E661E2C346A3354EBA5599321777B00E3DA0D3B822A

Re-scanning graph post-heal to verify loop resolution...
STATUS: Graph is healthy. 0 loops remaining.

---

License

This project is licensed under the MIT License - see the LICENSE file for details.