Probe mode
Direct port-forwarding, DDNS and public reverse-proxy paths can be tested from outside. Success means verified TCP reachability.
Measure → diagnose → fix → retest
Find what is actually wrong with your connection.
Measure responsiveness under load, verify a public TCP endpoint from outside your network, or diagnose why Plex, a NAS, a home server or Minecraft Java cannot be reached.
No account required. Connection Vitals separates measured evidence from guidance and never claims to have tested a path it cannot probe.
TCP reachabilityCGNAT / double NATLoaded latencyPlex / NAS / home server
Example remote-access result
Direct Plex path not reachable
Local PlexWorking
TCP 32400Not reachable
Router WAN IPv4100.72.x.x
DiagnosisCGNAT indicated
Next movePublic IPv4 or relay
Choose the problem
One diagnostic engine, different entry points.
The pages below are not separate thin calculators. They feed use-case context into the same measurement and decision engine.
Port Checker
Test one TCP port from outside and continue into forwarding/NAT diagnosis.
Check a port →Plex Remote Access
Separate Direct from Relay, verify TCP 32400 and check upstream NAT.
Diagnose Plex →NAS Remote Access
Direct, DDNS, reverse proxy, VPN or vendor relay — test the correct path.
Diagnose NAS →CGNAT Check
Compare router WAN IPv4 with the public IPv4 instead of guessing from a failed port.
Check CGNAT →Minecraft Java
Test TCP 25565 and follow the same forwarding/firewall/NAT diagnosis.
Test Java server →Bufferbloat
Measure idle vs loaded latency and compare before/after network changes.
Run responsiveness test →Measurement honesty
Not every connection method should have an open public port.
Guidance mode
Tailscale, QuickConnect and other relay/overlay paths may correctly have no public inbound port. We do not force those into a fake open/closed result.
Application layer
For selected HTTP-based services, a reachable socket can be followed by a lightweight service response check.
Responsiveness
Why does fast internet still lag?
Use the existing bufferbloat engine to measure idle, download-loaded and upload-loaded latency, then compare a fix with your baseline.
Remote access
Why can’t the internet reach my server?
Use TCP reachability, WAN/public IPv4 comparison and use-case context to separate application, firewall, router and upstream NAT problems.
Not sure which metric explains the problem?
Start with Internet quality: latency, jitter, loaded latency and throughput if you need to identify the signal before running another test.