TCP Port Checker

Enter the domain of a system you own or have permission to test. Do not enter a local router address or a private machine name. A public port check observes the route to the resolved public host, which can include a proxy or load balancer rather than the internal application server.

Enter exactly one port number from 1 through 65535. The default is 443, commonly used for HTTPS, but choosing that number does not itself establish that the responding service speaks HTTPS. Port and application protocol are related configuration choices, not interchangeable facts.

Consent to the external measurement and run the check. The tool requests a short TCP sample from one probe. Read the actual returned probe location and network. A firewall can allow one origin and deny another, so the observation should not be generalized to all clients.

If a TCP response is observed, test the application separately when appropriate. A connection can be accepted while authentication, TLS or the application itself fails. If no successful response is reported, retain the raw output and investigate authorized configuration; do not immediately conclude that a router forwarding rule is the sole cause.

A port checker answers a transport-level question about a specific target and port. It does not identify all services on a machine, guarantee that a process is healthy or test UDP behavior. This component uses Globalping’s TCP ping mode with three requested packets and shows the provider’s counters. A positive response is useful evidence from that route. A timeout remains ambiguous because packet filtering, routing, target state or the remote service can prevent a response. The page deliberately avoids turning every non-response into a confident “closed” label.

Do not use the port checker to enumerate third-party infrastructure without permission. There is no range scanner, parallel target list or background retry loop. The bounded tool is intended for checking a known service you administer.

Testing a domain behind a CDN can reach the CDN edge rather than your origin server. The port result then describes that public endpoint. Keep the resolved target and network architecture in mind when troubleshooting.

A successful TCP check does not validate a certificate or complete an authenticated transaction. Use the TLS checker for the certificate observation and your authorized application tests for the actual user operation.

Suppose the TCP sample to port 443 receives replies but the HTTPS check fails certificate validation. The transport observation and the TLS observation can both be correct: a responding port does not imply that the certificate is trusted. In another case, a known service may permit only a specific office network, so a remote public probe receives no response while authorized clients work. That is why the port checker retains origin and raw output. It describes what the probe observed, not the complete firewall policy or a list of every permitted client.

Keep the question, input, observation time and source together. A bare address, percentage or status code is difficult to interpret later because it omits the connection and measurement context. When comparing two observations, note what changed between them: the target, network, VPN state, selected adapter, record type or remote origin. Change one relevant condition at a time when possible. That makes the comparison easier to explain without inventing a cause from one number.

Save only information you actually need. Network records can reveal infrastructure names and connection details, even when they do not contain a password. Remove unrelated identifiers before sharing a screenshot with another person. Never put private credentials in a domain field or publish an internal hostname as part of a public test. A copied result is a snapshot, not a promise that the same conditions still apply when someone reads it later.

Kepo is a Mac desktop-widget product for repeated checks and small workflows. A labeled troubleshooting note or a supported monitoring widget can keep the relevant context accessible while you work. The download link opens the product workflow; this article does not automatically install a configured diagnostic widget, transfer your network inputs or start a scheduled check. Confirm the actual widget’s capabilities before relying on it for ongoing observations.