Global Network Health
Component Status
Live Traffic Analysis
01. Network Architecture Report
The Torzon infrastructure is built on a "Mesh of Trust" topology designed to eliminate single points of failure (SPOF). Unlike traditional centralized marketplaces that rely on a single database server, Torzon operates a distributed ledger for order tracking, synchronized across three distinct geographical locations (cold sites).
Our frontend uses Nginx with a custom Lua module dubbed "Endgame". This module inspects Tor circuit headers. It identifies patterns associated with DDoS attacks—such as "Circuit Cannon" or "Slowloris"—and drops the connection at the socket layer before it consumes PHP resources. This is why you might see a "Checking your browser..." screen during high-alert periods.
Latency & Circuit Health
Latency on the darknet is inevitable due to the triple-hop routing nature of Tor. However, we actively monitor the congestion of "Introduction Points". The table below shows estimated latency based on your entry node region. If you experience timeouts, we recommend forcing a new Tor identity.
| Region (Entry Node) | Avg Latency | Packet Loss | Status |
|---|---|---|---|
| North America (US/CA) | 120ms | 0.5% | Optimal |
| Europe (DE/NL/FR) | 95ms | 0.2% | Optimal |
| Asia Pacific (JP/AU) | 280ms | 1.5% | Congested |
| Eastern Europe (RU/UA) | 190ms | 0.8% | Stable |
02. Incident Post-Mortems
Transparency is the currency of trust. Below are the detailed technical breakdowns of recent service interruptions.
Incident #2025-003: Mirror Pool Desync
Root Cause: A scheduled database schema update caused a lock contention on the primary table `orders_active`. This prevented the secondary mirrors from reading new order statuses, resulting in users seeing outdated "Pending" statuses on deposits.
Resolution: The write-master was temporarily isolated. We rolled back the schema change and applied a hotfix to index the `transaction_id` column. Sync was restored at 14:45 UTC. No funds were lost, but 120 deposits required manual rescan.
Incident #2025-002: Layer 7 Flood (Botnet)
Root Cause: A sophisticated botnet (approx 15,000 IPs) targeted the `/captcha` endpoint, attempting to exhaust CPU resources by generating millions of image challenges per second.
Resolution: The Endgame filter was elevated to DEFCON 2. We enabled "Proof-of-Work" (SHA-256 client puzzle) for all incoming connections. This successfully filtered out 99.9% of the bot traffic as the bots could not compute the puzzle hashes fast enough. Legitimate users experienced a 5-second initial delay.
Incident #2024-128: Monero Node Fork
Root Cause: Our primary XMR node stuck on a minor fork of the blockchain due to a network partition in the Tor network.
Resolution: The system automatically failed over to the backup Monero node (Node B) within 15 minutes. However, deposits made during the window appeared unconfirmed until Node A was manually re-synced.
Client-Side Optimization Guide
Experiencing connection issues? 90% of "Downtime" reports are actually client-side misconfigurations. Follow these steps for maximum stability:
-
1. Disable JavaScript (Safest Mode):
While Torzon works with JS enabled, turning it off (Shield Icon -> Safest) significantly reduces the attack surface and prevents many browser-based timeouts. -
2. Use "New Identity" vs "New Circuit":
"New Circuit" only changes the route for the current tab. "New Identity" restarts the browser and clears all session data. Use "New Identity" if you are stuck in a captcha loop. -
3. Check Time Synchronization:
Tor encryption relies heavily on system time. If your clock is off by even 5 minutes, you will see generic connection errors.