Understanding Nexus mirror architecture
Nexus runs four separate .onion addresses simultaneously. They share the same backend database, the same user accounts, and the same order history. The only difference between them is the entry point. Understanding how this works helps you troubleshoot connection issues and avoid falling for "mirror down" social engineering used by phishing operators.
Why multiple mirrors exist
DDoS attacks are a constant reality for darknet platforms. A single address is a single point of failure — one coordinated flood of traffic and the entire platform becomes unreachable for hours. Nexus addresses this with four mirrors that distribute incoming connections across separate Tor hidden service relays. When one mirror is hit hard enough to slow down, the others absorb the traffic. In practice, a sustained DDoS event degrades one entry point while the remaining three continue handling requests normally.
There's a secondary reason. Tor's hidden service infrastructure occasionally loses sync on specific circuits — a node failure in the rendezvous chain, not a platform issue. A mirror switch gives your Tor client a fresh circuit attempt, which resolves connection issues that have nothing to do with the platform itself. Don't debug for twenty minutes when switching the address solves it in thirty seconds.
How traffic is balanced
The four mirrors are not load-balanced in the traditional sense. There's no central router deciding which mirror receives your request. Each mirror is an independent Tor hidden service with its own descriptor. Your Tor client independently establishes a circuit to whichever address you paste in. Different users naturally connect to different mirrors based on which circuit their Tor node builds first — passive distribution that protects against volumetric attacks targeting a single descriptor.
When you copy a mirror link and paste it into Tor Browser, you get a fresh six-hop circuit to that specific hidden service. Circuit quality varies based on which Tor relays are in your path that session. A slow connection on mirror three might resolve instantly on mirror two — same platform, different circuit geometry. This is normal. Try the next address before concluding the platform is having issues.
When a mirror goes down
Don't wait. Close the connection, copy the next address from this page, and reconnect. Your account state is stored server-side — balances, active orders, and messages are identical across all four mirrors. Nothing resets when you switch. You're not losing a session.
If two or three mirrors simultaneously fail to load, check the Nexus announcement thread on Dread for admin updates. Large-scale outages are logged there within hours. Combined uptime since launch has held at 98.1% across all four nodes — simultaneous total failure is rare, and when it happens it's typically brief scheduled maintenance rather than a platform emergency.