Does the name resolve correctly?
Start here when a hostname is involved.
Compare the returned address with the endpoint you actually expect. A successful DNS lookup only proves name resolution, not service reachability.
A field guide for isolating ordinary network, server and web failures without guessing. Work from name resolution through routing, TCP, TLS and HTTP, then inspect the host and firewall where the evidence points.
Start here when a hostname is involved.
Compare the returned address with the endpoint you actually expect. A successful DNS lookup only proves name resolution, not service reachability.
Separate local service state from remote reachability.
Refused often means the destination responded but nothing accepted that port. Timeout more often points toward routing, filtering, or an unresponsive host.
Certificate, SNI and chain diagnostics.
Always send the expected hostname with -servername when testing virtual hosts; otherwise you may inspect the wrong certificate.
Reachability is not the same as application health.
Timing separates slow DNS, connection establishment, TLS negotiation and server response.
Use tight filters and short captures.
Capture only traffic relevant to the issue and handle packet captures as potentially sensitive operational data.
Determine what policy exists first.
Do not disable a firewall as a first troubleshooting step. Confirm the service is listening, identify the path, then inspect the relevant rule or policy.
Walk the stack in order.
curl -v and note status/redirects.Find which boundary drops the connection.
Resolver vs authoritative data.
dig +trace to inspect delegation toward authoritative servers.Expiry, hostname and trust chain.
openssl s_client.Local routing evidence.
The return path matters too. A valid outbound route does not guarantee replies can get back through asymmetric routing or policy.
Map network symptoms back to a process.