Status dashboard FAQ
How the numbers are measured, where they come from, and how to read them.
Where does the uptime data come from?
A distributed prober fleet checks every TorZon Market mirror every 5 minutes via fresh Tor circuits. Success is a 200 response inside 60 seconds. Failures are re-checked from a different exit before being counted, to filter noise from a single bad circuit.
What is p50 and p95 latency?
Median (p50) and 95th percentile (p95) time-to-first-byte across the last 30 days of successful checks. p50 shows the typical experience. p95 shows the tail, what a bad circuit looks like when it still works.
Why does mirror age matter?
Older mirrors carry accumulated reputation with Tor relays and tend to route via known good guards. Freshly rotated mirrors sometimes bounce around for a day or two while descriptors propagate. The dashboard shows age so you can weigh that.
MTTR and MTBF?
MTTR is mean time to recovery, average length of a downtime window. MTBF is mean time between failures, average interval between outages. Both are aggregated over 30-day windows.
How is a mirror "down"?
Three consecutive check failures within 15 minutes from at least two different prober regions count as down. One-off timeouts are logged but not counted as downtime.
Is this data public elsewhere?
Not currently. The raw check log is retained for 90 days for internal analysis. The dashboard shows aggregated metrics only.
Which mirror should I pick?
Highest uptime and lowest p95 latency for a low-friction session. On the comparison page the top row is the current recommendation, sorted by a weighted score of uptime plus (1/latency).
Does the dashboard replace signature verification?
No. Uptime tells you the mirror was reachable, not that the address is the real TorZon operator address. Always verify the signed rotation first, use uptime as a tie-breaker second.